The FBI is investigating an attack on its own systems after ShinyHunters claimed responsibility for the incident, putting the prolific cybercrime group in the most direct conflict yet with agents responsible for investigating data extortion attacks.

The Monday breach, first reported by 404 Media, allowed ShinyHunters to temporarily deface the FBI jobs site. The group claimed it stole “very sensitive data on almost all FBI agents and individuals who filed an application with the FBI for a job,” in a lengthy post on its data-leak site.

“The FBI is aware of claims regarding unauthorized activity affecting FBIjobs.gov and is currently investigating,” a spokesperson for the agency said in a statement.

An alert on the FBI jobs site notes that apply.fbijobs.gov and the Special Agent Application Portal are currently unavailable.

The attack marks a sobering escalation by ShinyHunters, a notorious group that previously targeted major cloud platforms, healthcare organizations, universities, technology companies, retailers and education service providers. Previous victims of ShinyHunters this year include Instructure, Salesforce, Snowflake and McKesson.

“The ShinyHunters ransomware group appears to be actively trying to put a target on their back,” Cynthia Kaiser, senior vice president at Halcyon’s ransomware research center, told CyberScoop. 

ShinyHunters claims it targeted the FBI in response to a public service announcement it says contains false allegations about the group. The FBI issued the PSA following ShinyHunters’ May attack on Instructure, the company behind Canvas, a widely used central hub for K-12 and university coursework, exams and communication. 

The group responded with its own “PSA” on its data-leak site, insisting it is not affiliated with The Com, has never conducted swatting attacks or claimed it had sensitive or compromising information, including embarrassing photos or videos, to extort victims. 

The PSA was addressed to Brett Leatherman, assistant director of the FBI’s cyber division, and FBI Director Kash Patel. 

“While ShinyHunters has in the past been hyperbolic about the criticality of the data they’ve accessed, the group has established itself as a legitimate threat,” Flashpoint analysts told CyberScoop. 

“This attack benefits ShinyHunters by bolstering their reputation as a credible threat,” the analysts added. “In the group’s statement on their leak site regarding the breach, they portray the FBI’s PSA as an “attempt to ‘disrupt’ our operations and hinder clients’ trust in our organization hoping nobody pays us.”

The threat group typically uses social engineering, abuses weaknesses in identity systems or exploits vulnerabilities to gain access to cloud-hosted environments containing troves of sensitive or proprietary data, which it threatens to leak if the victim doesn’t pay a ransom.

ShinyHunters doesn’t appear to be seeking a payoff in this case, but rather a bid to coerce the FBI into amending or removing the May PSA. The group didn’t make any direct threat in the data-leak site post to release the stolen data, but it set a deadline of one week for action.

That coercive approach toward the FBI could backfire, according to experts. 

“Ransomware groups are largely successful because they operate like businesses,” said Kaiser, a former deputy assistant in the FBI’s cyber division. “Targeting other criminal groups or law enforcement — especially in ways intended to publicly shame — demonstrates a lack of discipline that historically has led to takedowns, takeovers or defections.”

The post ShinyHunters claims attack on FBI exposes almost all agents appeared first on CyberScoop.

Microsoft, along with a group of industry partners, disrupted EvilTokens, a short-lived but highly consequential cybercrime platform that investigators linked to more than 12,000 compromised Microsoft customer email inboxes across more than 10,000 organizations globally, the company said Tuesday.

Acting on federal court order Sept. 15, Microsoft and partners seized 50 websites the phishing-as-a-service used for operations and disabled more than 175 domains linked to EvilTokens’ supporting infrastructure. 

EvilTokens, launched in February 2026, was “a powerful cybercrime platform that used AI at every step of the attack chain — from compromising email accounts to designing intricate roadmaps for financial fraud and scams,” Steven Masada, associate general counsel and general manager of Microsoft’s Digital Crimes Unit, wrote in a blog post.

About 1,000 cybercriminals used EvilTokens over the course of its operation, a Microsoft spokesperson told CyberScoop.

The service was centered on an AI-style chatbot that cybercriminals used to analyze victims’ inboxes, identify trusted relationships, payment authorizations and other sensitive details that could facilitate fraud.

“AI was not simply helping attackers write more convincing messages. It helped them decide who to target, who to impersonate, and how to most effectively exploit the relationship to extract as much money as possible,” Masada wrote. 

EvilTokens was one of the most widely used phishing-as-a-service platforms prior to its takedown. It facilitated business-email compromise campaigns by stealing session tokens that allowed cybercriminals to sift through a victim’s inbox and maintain persistent access.

“We cannot estimate the total fraud attributable to all EvilTokens activity. However, we were able to correlate at least 13 complaints filed with the FBI’s Internet Crime Complaint Center to EvilTokens-linked activity, representing approximately $1.7 million in reported losses,” a Microsoft spokesperson said. “Because many incidents go unreported and not all victims can be definitively linked to specific campaigns, we believe this is a conservative estimate.”

Victims of EvilTokens were largely concentrated in the United States, Canada, the United Kingdom, Australia, India and France, according to Microsoft. SpyCloud, which supported the takedown, identified compromised email domains spanning 79 countries.

Microsoft said it also identified two men behind EvilTokens — Felix Utomi and Waidi Segun Adams — and attributes the development and support of the platform to Storm-2992, a threat actor unaffiliated with any other known cybercrime groups.

The United Kingdom’s Metropolitan Police acted on that information Sept. 18 when it served warrants in the greater London area, arrested the men accused of making articles for use in fraud and money laundering and seized their digital devices.

The Metropolitan Police said it received information from Microsoft about EvilTokens’ administrators in August. Utomi and Adams were released on bail as the investigation continues. 

“The two primary operators identified in our investigation were residing in the U.K.,” a spokesperson for Microsoft told CyberScoop. “While our investigation focused on those individuals, we believe others may have supported the operation in various capacities.”

Microsoft’s legal filing in the U.S. District Court for the Eastern District of Virginia refers to five additional unidentified people allegedly acting as support personnel and users.

Microsoft and others involved in the EvilTokens takedown, including Health-ISAC, Cloudflare, OpenAI, Shadowserver and TRM Labs, didn’t fully quantify how much fraud the service enabled, but it gained popularity quickly among cybercriminals and was lucrative for its operators.

Coinbase, which also aided the investigation into EvilTokens, said it traced about $1.1 million in revenue for EvilTokens from its paying customers. The virtual currency company’s threat researchers found more than 1,000 deposits to EvilTokens from more than 700 distinct addresses through June 2026. 

Operators sold access to the service through Telegram for a $1,500 initiation fee and a recurring $500 subscription. EvilTokens significantly lowered the barrier to entry for cybercriminals by including specialized tools for identity attacks, cloud systems, social engineering and financial fraud in a single interface.

The service allowed cybercriminals to map organizational structure and permissions in Microsoft Graph, which enabled lateral movement, researchers said. With active tokens gained through a collection of highly-targeted phishing lures, cybercriminals consistently bypassed multi-factor authentication, email gateways and endpoint security tools.

Microsoft said the platform’s creators developed portions of the platform with AI and it uncovered capabilities from multiple AI models. 

“It packaged much of the criminal process into a commercially run service, complete with subscription pricing, customer support, management dashboards and tools designed to move customers from account access toward financial exploitation,” Masada added.

The companies and organizations involved in the globally-coordinated takedown identified and notified potential victims, shared indicators of compromise and shared intelligence with law enforcement about EvilToken’s operators and some of its customers.

Experts advised organizations and employees to treat unsolicited device codes as a red flag, assume compromised accounts are fully cataloged in minutes, and independently verify requests to change payment information or redirect funds.

“The infrastructure supporting EvilTokens has been disrupted, but the model it demonstrated will not disappear with it,” Masada warned.

The post Microsoft and partners disrupt EvilTokens, a comprehensive cybercrime service for financial fraud appeared first on CyberScoop.

Researchers said they used Anthropic’s Claude and OpenAI’s Codex to identify a damaging flaw embedded in a popular software decoding tool that could leave major internet platforms, enterprise services, and web frameworks vulnerable to data theft and remote access.

The vulnerability, nicknamed HEIF Heist, refers to the malware’s ability to trigger memory corruption errors in affected software, allowing the attacker to pilfer sensitive data from its victims. In a report published Thursday, the researchers laid out the potential damage an attacker could cause, including gaining access to internal OpenAI repositories, leaking user files, access tokens, and other sensitive data for online services like Amazon Web Services, and gaining remote code execution privileges across a range of online services, including Meta’s core product suite, GitHub Enterprise servers and open-source internet forum Discourse.

“Even when Remote Code Execution isn’t immediately achievable, the attack primitives may still allow arbitrary heap disclosure, letting an attacker ‘heist’ in-memory data such as other users’ data and environment variables,” wrote Hacktron researchers Harsh Jaiswal, Mohan SRK, Rahul Maini and Sudhanshu Rajbhar.

The researchers relied heavily on AI systems, including frontier models from OpenAI and Anthropic, to conduct their research. Attribution for the research is described as being “led” by the Hacktron human researchers “assisted by Hacktron Harness, GPT-5.6 Sol, and Opus 5.”

According to the research, the attack exploited the way that code parsing tools in many popular software decoders — specifically libheif and libde265, used to parse C and C++ software — process certain image files.

By uploading HEIF, HEIC and AVIF image files corrupted with malicious code, the attacker could bypass most of the victim’s application layer defenses, in many cases achieving remote code execution privileges for accounts or products tied to major AI and tech brands.   

While the latest version of libheif has been patched, the researchers said “any deployment lacking the latest upstream security patches is potentially vulnerable.”

In one incident detailed in a Sept. 13 blog, Jaiswal, Maini, and Hacktron researcher Mohan Pedhapati described how chaining two vulnerabilities, including an image parser flaw, could compromise OpenAI employee accounts.

With access to the compromised accounts, researchers could reach OpenAI’s internal repositories. As a proof of concept, they opened a pull request in the company’s “monorepo,” a centralized library where code is shared across projects, using the employee’s Codex credentials. 

According to a timeline provided by the researchers, the flaw was discovered on July 25 and patched within days. They said the entire attack, from discovering the initial vulnerability to gaining access to the repositories, took less than 72 hours. OpenAI paid them a bug bounty of $6,500 for their work.

Given that AI models are increasingly integrated into enterprise and personal networks, an attacker exploiting HEIF Heist could have accessed far more than just OpenAI’s systems and data.

“Until two months ago, a user or OpenAI employee logging into OpenAI’s own help forum could have had their ChatGPT and Codex accounts taken over,” the researchers wrote. “Since people can connect various services to Codex and ChatGPT, the scope of what we could theoretically access was huge, including GitHub, Slack and emails.”

CyberScoop has reached out to OpenAI for comment on the research and additional information.

At the same time, the researchers said the attack paths they found were not particularly easy or efficient to exploit.

“Exploitation requires fingerprinting the target version and tailoring the payload images,” the blog stated. “Some of our RCE attempts landed only after thousands of image uploads. That said, an AI agentic approach with a frontier model like GPT-5.6 Sol cut exploit development time down to roughly 1 to 3 days from initial probe to remote RCE. A motivated attacker can convert a vulnerable upload endpoint into RCE or an info leak.”

The post Researchers use AI to find widespread software decoder flaw  appeared first on CyberScoop.

The past few weeks have “felt very strange” for Juan Andres Guerrero-Saade.

Like many, he is trying to sort through the spate of frontier-model AI agents from OpenAI, Anthropic, Meta and others hacking their way onto the open internet over the past few months, particularly amid the already-heated national debate around the emerging technology and its impact on society.

Guerrero-Saade, a fellow for AI and security research at SentinelOne and an adjunct professor at Johns Hopkins University, said the hacks are worth taking seriously, but at a time when businesses and open-source maintainers should be focused on further hardening their systems and policymakers should be discussing new solutions,  “what we see is cybersecurity being used essentially as an excuse for these AI doomer arguments.”

The incidents have spawned those “doomer arguments” amid an intense public debate about the technology, the pace of industry development, and whether government and the private sector are doing enough to protect against “doomsday”-type scenarios, where AI systems take over or attack large parts of the internet or society.

Guerrero-Saade is among a growing chorus of cybersecurity professionals who say that while AI systems pose real, unique threats to our systems, the apocalypse is far from inevitable. Most of the public concerns around the incidents, let alone worries about killer AIs attacking critical infrastructure, assuming control of the internet and wiping out humanity, are either technically impossible or can largely be controlled through established cybersecurity principles.

There is this “narrative or magical thinking of ‘Well, AI is going to be able to hack everything, and therefore it can control everything, and therefore it’s going to kill us all,’” he told CyberScoop. “And you [think] these just don’t add up. They’re not very well-reasoned arguments.”

This fatalistic narrative tied to AI’s eventual dominance doesn’t hold up under scrutiny, according to experts CyberScoop spoke with. In recent conversations, cybersecurity and national security professionals raised questions about both the technical solutions OpenAI and Anthropic use to contain their models, as well as the glaring absence of federal oversight from federal regulators or truly independent third-party review.

For example, Jacob Coxon, an Anthropic employee who resigned over AI safety concerns, told CBS News that frontier models could not be “unplugged” by humans once deployed because the model would copy itself to thousands of other computers connected to the internet.

By contrast, Matt Tait, a former information security specialist at UK signals intelligence agency Government Communications Headquarters (GCHQ), pointed out that the models run by Anthropic and other frontier companies require extremely expensive, “ultraspecialist” machines that “are functionally supercomputers.”

“There is a zero chance that Anthropic’s most capable models will be able to extract their own model and run in the wild, because those supercomputers essentially only exist in datacenters,” Tait said.

“Not a credible warning”

Other former cybersecurity government leaders say the agentic hacks represent a failure by regulators and industry to deploy known technical and policy options that make it harder for these types of incidents to occur.

Matt Hartman, former deputy executive assistant director for cybersecurity at the Cybersecurity and Infrastructure Security Agency, said “we should not accept harmful AI behavior as inevitable or unmanageable.”

“There are meaningful steps companies can take to monitor agent activity, constrain permissions, detect anomalous behavior, and build stronger safeguards into how these systems operate,” said Hartman, now a chief strategy officer at Merlin Group. “Those controls will inevitably involve trade-offs in capability and speed, but that’s a familiar cybersecurity challenge. Our goal should be to manage the risk without unnecessarily limiting the enormous benefits AI can provide.”

Ciaran Martin, former head of the UK’s National Cyber Security Centre, took issue with the way the CEOs of frontier AI companies have framed the threat of “rogue” AI behavior as inevitable, while issuing dire warnings about future threats and capabilities with little transparency.

Martin’s comments came after an essay published by Anthropic CEO Dario Amodei that cited the threat of a HuggingFace-style swarm of agents that could create a botnet capable of “taking over the entire internet” within 6-12 months.

This, Martin said, “is not a credible warning,” because it doesn’t explain how the exploitation would function, how such a botnet would persist on the internet, or how it would escape law enforcement. 

 “It assumes no monitoring of systems, no anti-virus, no DDoS protection, no network segmentation, no incident management, no nothing of any kind of the cybersecurity on the global Internet of the type that has developed over the last 30 years,” wrote Martin. “For a claim of this magnitude, there is neither evidence for the contention nor a credible account of a path to this outcome.”

Meanwhile, some federal government cybersecurity leaders have touted the technology’s disruptive potential and called for more widespread adoption of AI tools by defenders.

Joseph Alm, assistant secretary of cyber, infrastructure and risk resilience at the Department of Homeland Security, said classified systems may retain stronger protections. But for most other data, AI models are “just going to know things and be able to infer things about the world, and we’re going to have to adapt to that as almost inevitable.”

Asked by CyberScoop whether the government or frontier AI companies could be doing more to prevent or deter their models from carrying out unauthorized hacks via agents, Alm cited recent efforts by the Trump administration this year to establish pre-release testing of commercial models as a step in the right direction. But he called unauthorized AI agent hacks “a new threat class” that is different from previous threats and can be easily distributed to users through open-source software today.

“I think what we can do is…encourage the building of good sandboxes, so that the best models aren’t used for this and the stuff you see out in the wild is the kind of detritus that you can actually respond to effectively and control your networks,” said Alm.

Other experts have shared similar concerns. Earlier this month, CrowdStrike CEO George Kurtz recently warned of a new threat class emerging alongside nation-states, cybercriminals, and hacktivists: “the agent state.” By pairing AI systems with small human teams, these operators can now match the speed, scale, and sophistication of government-backed hackers.

“It took a nation to fund the talent, the tooling, the infrastructure, the patience,” said Kurtz. “That scarcity is over.” 

To be sure, frontier AI companies tout their commitment to both approaches. OpenAI and Anthropic have rolled out an array of cybersecurity partnerships, external red-teaming programs, vulnerability disclosure programs and cybersecurity technical advisory bodies filled with cybersecurity experts.

Mohammed Husain, strategic delivery lead for government at OpenAI, told CyberScoop that the company deploys both internal safety guardrails for their models and relies on outside cybersecurity vendors for additional expertise.

Internally, OpenAI focuses on vulnerabilities at the training level: filtering data poisoning attacks, blocking harmful datasets, and using network controls to prevent prompt injections. For other security layers like sandboxing, identity management, networking controls, they outsource to external vendors. 

“I don’t think OpenAI has all the answers here but what we do as a research lab is we’re going to focus on levels of protection we have expertise in and we partner to self-complement,” said Husain.

AI safety vs. AI cybersecurity

In response to the HuggingFace hack, OpenAI and Anthropic have allowed third-party organizations, such as nonprofit AI research firms METR and Redwood Research, to investigate. But multiple cybersecurity professionals told CyberScoop that both firms lack incident response experience and focus primarily on AI alignment and safety. Their reporting on the hack also lacked critical details: network monitoring logs, telemetry, and other data standard in cybersecurity threat intelligence reports.  

METR president Chris Painter addressed those general concerns in a post on X, saying since 2022 the organization has worked with Google, Anthropic, OpenAI, Meta, Amazon and others on investigations and third-party evaluations. Painter said none of the AI companies fund METR and that his employees are not uniformly “doomer” or “accelerationist” around AI.

Painter also said METR’s work ensures that if AI systems become autonomous or “rogue” within a company, there are ways to share that information with governments and people “outside the company’s walls.”

“We don’t accept money from frontier AI companies,” wrote Painter. “They haven’t paid us for our work, and we don’t accept donations from them or their employees. As we’ve shared previously, multiple frontier AI companies currently provide us with free access to their models in order to perform our evaluations, research, and engineering.”

AI safety and AI cybersecurity advocates take different approaches to securing “rogue” AI behavior. Safety advocates focus on aligning models around ethical training and behavior. Cybersecurity advocates argue that technical and regulatory controls must go further—actively preventing models from accessing what they need to carry out malicious behavior.

Guerrero-Saade said sandboxes in particular can easily be programmed with aggressive cybersecurity monitoring in order to spot when something odd may be happening and react in real time.

“I can’t think of an easier situation in which to set up trip wires, set up configurations like DNS servers, just different parts where you can say ‘Hey, anomalous behavior is happening,’” he said. “We should have been able to tell this immediately, not weeks and months later. So watching [the AI hacking incidents] go down is a little ‘crazy-making’ because we’re seeing things that, frankly, look like neglect, negligence, people just mishandling things, and then being told that these are categorically new incidents that mean that AI systems need to be treated completely different from anything that’s come before.”

While cybersecurity experts say AI systems are, at their core, still software, they do operate differently from more traditional code in ways that can make them harder to predict and control.

John Hultquist, chief analyst at Google’s Threat Intelligence Group, said most software has been deterministic. It may have bugs or vulnerabilities, but an expert could generally understand how it would react to certain stimuli, making it easier to design straightforward controls.

AI models are non-deterministic, with far more variability than traditional software. That can break security controls that rely too much on predicting behavior in advance. Using AI to enforce security controls on other AI models faces the same problem: the systems being deployed to control AI are just as unpredictable. 

But people are also non-deterministic, and people have developed systems in other industries and practices to account for that.

Hultquist drew on his Army experience, noting that “they give incredibly dangerous, expensive things to 18-year-olds” and expect responsible use. The military manages this through two types of controls: deterministic ones like strict weapons and ammunition protocols, and non-deterministic ones like human officers who monitor and correct violations.

Similarly, established cybersecurity controls have been used by incident responders to detect and prevent or mitigate ongoing cybersecurity breaches.

“I don’t think we should throw out all the other tools that we have learned to use as well. I think that would be utterly foolish,” he said, later adding “I will say that if we use only non-deterministic tools to figure out when things are happening, we shouldn’t be surprised when we get the wrong answer.”

The post The AI hacking apocalypse is not inevitable appeared first on CyberScoop.

Attackers of unknown origins and motivations are exploiting a critical zero-day vulnerability in Cisco Secure Email Gateway, authorities and researchers said Monday.

The vulnerability — CVE-2026-76461 —  was exploited before Cisco disclosed and patched the defect Monday and allows unauthenticated, remote attackers to execute commands with root privileges on vulnerable systems. “In practical terms, that gives the attacker control of the gateway itself,” Douglas McKee, director of vulnerability intelligence at Rapid7, told CyberScoop.

Cisco said its product security incident response team became aware of active exploitation of the defect affecting Cisco AsyncOS Software for Cisco Secure Email Gateway in September. When asked for further details, a company spokesperson pointed to the advisory and reiterated that the company is aware of active exploitation of the vulnerability.

The company did not say how many organizations are impacted by active exploitation thus far, but it indicated multiple customers were likely compromised prior to disclosure. 

“Cisco has conducted a thorough threat intelligence investigation on devices that belong to Cisco Secure Email Cloud. Cisco has directly contacted customers who own Cisco Secure Email Cloud devices where indicators of possible compromise were identified,” the company wrote in its security advisory. “Cisco is engaged in remediation and recovery operations. Cisco has already deployed mitigations that are within Cisco’s management.”

The Cybersecurity and Infrastructure Security Agency added the zero-day, which affects cloud-based and on-premises instances of Cisco Secure Email Gateway, to its known exploited vulnerabilities catalog shortly after Cisco’s disclosure. 

The tight timeline between Cisco’s public advisory and patch guidance, and CISA’s quick addition to the KEV catalog indicates the vulnerability deserves immediate attention, McKee said. 

“The combination here is pretty ugly. No authentication is required, an attacker can reach the vulnerable code by sending an email through the appliance, successful exploitation can result in root-level command execution, and Cisco has observed exploitation in the wild,” he added.

Researchers at Rapid7 and VulnCheck said they don’t yet know how many organizations are impacted by active exploits, but they encouraged Cisco customers to patch and hunt for potential signs of compromise as soon as possible. 

Spencer McIntyre, director of exploit development at VulnCheck, told CyberScoop the exploit could allow an attacker to maintain access to the email gateway and monitor communications. “Stealing or silently snooping on email comms is a common tactic for state-sponsored and other threat actors conducting espionage operations,” he said. 

“It’s going to be worse for organizations that have the appliance deployed on-premises. In this case, the attacker could pivot internally,” McIntyre added. “If, however, organizations use a cloud instance, the compromised gateway is less likely to have significant access to internal organizational resources.”

Cisco released indicators of compromise to help customers hunt for attempted exploitation in their environments, but the company added that attackers could remove or hide those traces with the level of access granted via exploitation.

The post Cisco warns customers of actively exploited zero-day in email gateways appeared first on CyberScoop.

Proofpoint researchers have spotted at least four state-aligned threat groups chain a trio of zero-day vulnerabilities to conduct espionage on various targets of interest to China’s government since late August. 

The Chinese espionage group that Proofpoint tracks as TA412, also known as Violet Typhoon and APT31, struck first, exploiting the chain of vulnerabilities Aug. 28. At least three additional espionage threat groups followed suit, exploiting the same vulnerabilities in subsequent waves of attacks days later, researchers said.

The exploit chain Proofpoint calls BlueMoon targets Chrome, Chromium-based browsers and Microsoft Windows. It allows attackers to run code in the browser’s sandbox, escape the sandbox and gain system privileges to access a targeted machine, said Mark Kelly, staff threat researcher at Proofpoint.

“All three vulnerabilities were exploited before patches were available to the public,” he said.

The vulnerabilities include: CVE-2026-85046 and CVE-2026-87491, remote-code execution defects in the JavaScript engine for Chromium-based browsers; and CVE-2026-85880, a privilege-escalation zero-day that Microsoft disclosed Tuesday in Windows Advanced Local Procedure Call. 

“While the V8 vulnerabilities were known and fixed in Chromium source code, they were not yet patched in the latest publicly available browsers at the time of the activity, meaning they effectively functioned as zero-days in those products,” Kelly said.

Proofpoint said the exploit kit developer likely reverse engineered the publicly available Chromium patches to weaponize the browser exploit chain during that gap.

With a limited group of organizations exposed to all three vulnerabilities, attackers moved quickly and likely rushed development to target a narrow pool of potential targets. “In all observed cases, the infrastructure used for exploit delivery was created on the same day as — or in the days immediately preceding — the associated campaigns,” Proofpoint wrote in a threat intelligence report.

APT31, a group that’s committed espionage on behalf of China’s Ministry of State Security, including seven Chinese nationals indicted by the Justice Department in 2024, dropped various lures containing the exploit chain loader in phishing emails targeting non-governmental organizations, mining companies and commodity trading firms in the United States. 

The phishing link installed a malicious browser extension disguised as Google Gemini on targeted machines, enabling attackers to surveil browser activity, steal credentials and execute commands, according to Proofpoint. 

Other distinct threat groups have also used the BlueMoon exploit chain with some slight technical changes and variances in targeting. 

“Proofpoint observed BlueMoon usage as recently as Sept. 8,” Kelly said. “The activity peaked Sept. 2-3 immediately prior to the Chrome patch being released and has continued intermittently since then.”

A China-aligned espionage threat group Proofpoint tracks as UNK_LateNight targeted multiple U.S. aerospace companies Sept. 2. Researchers also that day observed UNK_DoubleCheck, a suspected espionage-motivated threat group targeting Vietnamese manufacturing organizations with emails from a compromised Southeast Asian government account. 

Researchers said UNK_QuietRacket, another espionage group aligned with China, targeted government, consulting and financial sector organizations in Indonesia and Singapore Sept. 3.

Proofpoint has directly observed fewer than 20 organizations targeted globally thus far, but Kelly said the true number of impacted organizations is likely much higher. 

While Proofpoint attributes most of the observed attacks to Chinese espionage groups, attackers of other origins and motivations could strike soon as well. 

“Given its ease of adoption, we expect the exploit kit is likely to proliferate further and be adopted by additional espionage-motivated and financially motivated threat actors as patched versions are fully rolled out across all Chromium-based browsers,” Kelly said.

The post Chinese espionage groups swarm to exploit triple-link chain of zero-days appeared first on CyberScoop.

Microsoft addressed 974 defects across its product suite, including two actively exploited zero-day vulnerabilities, in its monthly Patch Tuesday security program. 

The massive batch of patches, Microsoft’s largest ever, reflects a continuing trend for the vendor as it leans on artificial intelligence to discover more vulnerabilities at a faster rate. Yet, the recent period of record breaking vulnerability disclosures hasn’t resulted in a flood of actively exploited zero-days.

“AI-assisted vulnerability discovery shows no signs of slowing down,” Dustin Childs, head of threat awareness at Trend Micro’s Zero Day Initiative, wrote in a blog post Tuesday. “However, we have not seen a correlating spike in active exploits — yet.”

The vulnerabilities actively exploited prior to disclosure — CVE-2026-81963 affecting the Windows Update Stack and CVE-2026-85880 affecting Windows Advanced Local Procedure Call — both have CVSS ratings of 7.8 and allow attackers to escalate privileges. 

More than 1 in 10 defects Microsoft disclosed in this month’s security update are rated critical. The update included 723 vulnerabilities in Windows, 111 in Office, 111 in Office 2016, 62 in SQL and 22 spanning various developer tools.

Researchers encouraged security teams and customers to not get overwhelmed by the total number of defects, but instead focus on their specific areas of risk and exposure. 

“While the number of vulnerabilities being patched is rising, the number of vulnerabilities that can and will affect most organizations remains quite low. AI-assisted vulnerability discovery in 2026 is creating larger haystacks, but it isn’t finding more needles,” Satnam Narang, senior staff research engineer at Tenable, said in an email. 

“It’s critical that organizations understand which vulnerabilities actually apply to them, whether they pose a threat by being reachable and exploitable, and prioritize remediation based on this risk context,” he added. 

Jack Bicer, director of vulnerability research at Action1, drew a similar conclusion from the record-breaking Patch Tuesday. 

“At this scale, the challenge is not simply getting through the patch list but knowing what needs attention first,” he said. “With hundreds of updates landing at once, IT and security teams need to quickly separate the vulnerabilities that demand immediate action from those that can follow the normal deployment cycle.”

The full list of vulnerabilities addressed this month is available in Microsoft’s Security Response Center.

The post Microsoft discloses two actively exploited zero-days among 974 vulnerabilities appeared first on CyberScoop.

SonicWall customers are grappling with yet another pair of actively exploited zero-day vulnerabilities in SonicWall SMA 1000 appliances, a product that’s been besieged with recurring defects and attacks over the past nine months. 

The vendor disclosed and released patches for the defects — CVE-2026-83548 and CVE-2026-83549 — and noted both were already actively exploited in the wild in a security advisory Tuesday. The Cybersecurity and Infrastructure Security Agency added the defects to its known exploited vulnerabilities (KEV) catalog Wednesday. 

SonicWall customers have confronted a barrage of actively exploited vulnerabilities in SonicWall devices for years. Attackers have consistently exploited newly discovered zero-days and years-old defects in the vendor’s products to break into victim environments.

Rapid7 researchers said the new zero-days — a max-severity pre-authentication server-side request forgery vulnerability and a high-severity OS command injection vulnerability — can be chained together to achieve unauthenticated remote-code execution. 

SonicWall did not say how many customers have been directly impacted by active exploitation or when the first known instance of exploitation occurred. The company did not respond to a request for comment.

“Please stop us if you’ve heard this one before: Another appliance sitting at the edge of the network, another pair of vulnerabilities chained together, and another unauthenticated path to complete compromise,” Jake Knott, head of threat intelligence at watchTowr, said in an email. 

“SonicWall says these vulnerabilities were internally discovered, while also saying it investigated a case indicating active exploitation. Please pick one, or, at minimum, explain how both are true,” Knott added. “Those statements may be technically accurate, but without that context, the disclosure leaves defenders guessing about when and how the vulnerabilities were actually identified.”

The vendor’s security advisory did not include indicators of compromise. It urged customers to contact tech support for assistance in reviewing IOCs and hunting for potential signs of compromise, and if detected, to reimage or redeploy the appliance, change all user and administrator passwords and reset tokens. 

SonicWall did not attribute the known exploits to a specific threat group or describe the attacker’s motivations.

The freshly disclosed pair of vulnerabilities are the latest in a series of security issues confronting the vendor’s customers, including actively exploited zero-days, previously disclosed defects, and an attack last year that allowed a state-sponsored threat group to steal the firewall configurations of every SonicWall customer. 

In late July, Huntress researchers spotted an attack spree that compromised 30 SonicWall customers in less than two days. Earlier that month, the company acknowledged another pair of zero-days that were exploited for three weeks before the vendor disclosed and patched the defects.

Ransomware groups, including INC ransomware and Akira, have taken a special interest in SonicWall. Ten of the 19 SonicWall defects added to CISA’s KEV catalog since late 2021 are known to be used in ransomware campaigns.

The five defects added to CISA’s KEV most recently, since just mid-December 2025, all impact SonicWall SMA 1000 appliances.

The post Attackers exploit zero-days in consistently besieged SonicWall product appeared first on CyberScoop.

The FBI said a 17-year-old from Maine is the first child federally charged and adjudicated for crimes stemming from their involvement in 764, a violent extremist collective.

A judge ordered the teen to remain detained after determining they committed multiple crimes, including conspiracy to sexually exploit a child, sexually exploiting and enticing a child, distributing child sexual abuse material, sending interstate threats, cyberstalking victims and identity theft.

“This first-in-the-nation case should make it crystal clear that if you conspire to commit violent, extremist crimes, your age will not shield you from accountability,” Ted Docks, special agent in charge of the FBI’s Boston Division, said in a statement Tuesday. “What this juvenile did would shock most people to their very core, and it is our hope that by publicizing this case, others will be deterred from making the same devastating choices this teen did.”

The nihilistic extremist group the teen participated in, 764, is more broadly affiliated with The Com, a sprawling network of thousands of people, typically between 11 and 25 years old, seeking to foster social unrest by destroying civilized society through the corruption and exploitation of children and other vulnerable populations.

The Justice Department’s resolve in this case — detaining and adjudicating a 764 member before they reach adulthood — marks a turning point and apparent change in internal policy against charging children for federal crimes linked to their involvement in violent extremist groups. 

“Crimes from 764 copycat groups in The Com are extremely serious and it speaks to how law enforcement prioritizes these things,” Allison Nixon, chief research officer at Unit 221B, told CyberScoop. “They recognize this loophole involving minors needs to be closed in order to tackle this social problem of violence arising from minors which crosses state lines.”

The first-of-its-kind case has a wider impact that will cause ripples across the landscape of violent extremist crime, she added. 

“If you sexually exploit a child, the fact that you yourself are a minor will not protect you from the consequences of your actions,” Andrew Benson, U.S. attorney for the District of Maine, said in a statement.

The Maine teenager, whose identity is being withheld, was ordered to serve a term of official detention followed by supervision, the FBI said. Officials did not provide details about the terms of detention.

Andrew McCormack, assistant U.S. attorney for the U.S. District of Maine, and a spokesperson for the FBI Boston Division, both said federal law restricts what law enforcement can share about cases involving underage criminals and declined to say where the teen lived, where they’re being detained and for how long. 

Nixon conveyed, with some reluctance, the need for more actions like this targeting underage members of 764 and similar groups. 

“I’m not advocating normalizing throwing kids in prison but we very much need to find a new balance to protect society from violent groups that are incentivized by this federal loophole to commit maximum harm before turning 18, and then after 18, to recruit and train kids to do dirty work for them,” she said. “I cannot understate how much this loophole specifically influenced this culture of maximizing harm.”

The teen’s ordered detention marks a continuation of consistently heightened law enforcement activity targeting members of 764 and affiliated groups. 

Kyle William Spitze, an original member of 764 and leader of one of its offshoots, was sentenced to 77 years in prison, the longest imprisonment ever imposed on a nihilistic violent extremist, in federal court in Tennessee in late August.

Alexis Aldair Chavez, who began associating with 764 as a child in 2022 before leading an offshoot 8884, was sentenced to 40 years in prison in July for blackmailing and coercing multiple girls to commit self-harm, torture animals and degrade themselves on camera to produce CSAM. 

Other alleged 764 members arrested since 2025  include: Leonidas Varagiannis and Prasan Nepal, Baron Cain Martin, Tony Christopher Long, Erik Lee Madison, Zachary Sweeney and Aaron Corey.

The FBI said it is currently investigating more than 500 subjects nationwide who are allegedly involved in 764 and its many offshoots. 

“These groups actively target minors and are made up of a large percentage of minors. They are intentionally recruiting juveniles here in the U.S. to conduct criminal acts because, simply put, they think they can get away with it because historically, the federal justice system has rarely prosecuted juveniles,” Docks said. 

“We’re here to tell you, they’re wrong, and if they don’t stop this abhorrent behavior, they too could find the FBI on their doorstep and themselves in federal court. We are going to do everything in our power to protect kids and ensure those who harm them don’t get away with it,” Docks added.

Nixon, who has studied the rise of these violent extremist groups and helped law enforcement identify some of its members, said she’s pleased with this development. 

“Right now the prevailing sentiment within these violent groups is that you can do anything you want with no accountability before you turn 18, so therefore you should commit the most heinous acts possible because it’s your last chance. That’s why there are so many 17 year olds who become a major public nuisance, because it’s their last hurrah — but for bomb threatening schools and abusing little kids,” Nixon said. 

“They slow down on their 18th birthday and pivot to using minors to hide behind, and teaching them this lifestyle,” she added. “Closing that loophole will do a ton to break this cycle. They pay close attention to law enforcement, and just one arrest shatters their sense of safety. I don’t think the FBI will stop at just one arrest.”

The post Jail time for Maine child in 764 marks turning point in federal law enforcement appeared first on CyberScoop.

Sality, a Russia-based botnet that infected more than 11 million devices during a 23-year run of operations, was dismantled Monday by law enforcement, CrowdStrike and the Shadowserver Foundation. 

CrowdStrike, which announced the takedown Tuesday alongside authorities, said it played a crucial role dismantling the botnet’s technical infrastructure, rendering the malware-spreading operation irrecoverable. 

The peer-to-peer botnet was a persistent piece of criminal infrastructure that evaded disruption for an exceptionally long period because it lacked centralized architecture. 

Sality used infected machines to communicate peer-to-peer, creating a decentralized structure that made system-wide disruption efforts more difficult than botnets that rely on a core server. 

“The same properties that made Sality resilient also created the conditions for its undoing,” CrowdStrike wrote in a blog post. The company said it targeted Sality’s peer list of infected machines and tricked the network into permanently cutting off access to those devices.

“From the operator’s perspective, infected machines simply disappear,” CrowdStrike wrote, adding that the botnet is no longer under the operator’s control.

Sality’s domains were seized by a globally coordinated effort supported by the FBI, Justice Department and authorities from Europol, Bulgaria, Hungary and Romania, officials said. Shadowserver is working with internet service providers to identify devices infected by Sality and aid with remediation. 

“Cybercriminals, botnets, and malware are a clear and present danger to our nation’s security and economy,” Bill Essayli, first assistant U.S. attorney, said in a statement. 

Europol said the Sality takedown was the culmination of work spanning global law enforcement back to 2017. 

CrowdStrike said Sality’s operator was primarily financially motivated, but it attributed three DDoS attacks to Sality, suggesting the operator was occasionally willing to use the botnet for personal or political aims. 

The botnet enabled cryptocurrency theft and cyberattacks on victims in the United States and abroad, the Justice Department said. Officials did not name the person or cybercrime group behind Sality.

“This operation demonstrates that peer-to-peer architecture, long considered a shield against disruption, is not invincible,” CrowdStrike wrote. 

“Operating for decades without consequence does not mean operating without risk,” the company added. “The calculus has changed. We will find you, we will dismantle your infrastructure, and we will impose costs that make the enterprise untenable.”

The post Dogged Russia-based botnet dismantled after 23-year run appeared first on CyberScoop.

Unit 42’s top brass has seen enough from internal frontier AI model testing and malicious in-the-wild use of commercially available AI tools to be genuinely concerned.

“I can tell you without exaggeration that we believe that this is a generational shift in cybersecurity,” Sam Rubin, senior vice president of Palo Alto Networks’ threat intelligence arm, said in a media briefing Wednesday. 

A period of relative balance between security and exposure has been broken by frontier AI model capabilities that could allow attackers to find and exploit network weaknesses with speed, Rubin said.

Unit 42 warned that capabilities demonstrated by readily available agentic AI models, and those unlocked by frontier AI models that remain gated for defense, have shifted the balance of power from defenders to attackers.

“The defenses that we’ve had built up over years weren’t necessarily built for or prepared for these machine-speed attacks,” Rubin said. “Organizations are ill-equipped to detect and to respond quickly in the face of these attacks.

Back in April, when Anthropic brought Palo Alto Networks and other major technology companies together to form Project Glasswing, an initiative to find and address security defects with its Mythos model, Unit 42 estimated the same capabilities would be in the hands of attackers within a year. 

“Well, here we are five months later, and we’re starting to see the early waves of this threat in the wild,” Rubin said. 

Unit 42 is actively investigating an attack on one of its customers where an attacker used an agentic framework to exploit 50 applications and other weaknesses across the enterprise in less than 10 hours. Rubin estimates AI allowed the attacker to accomplish in 10 hours what would have taken at least 10 days in a pre-AI era. 

Attackers are already using AI across the entire attack chain, said Sherrod DeGrippo, vice president of threat intelligence at Unit 42. “We are not far from fully agentic attacks across all at once, but right now it’s piece by piece by piece,” she said. 

“AI has seeped into every part of what threat actors do,” including malware development at scale, delegation, social engineering and ransomware negotiations, DeGrippo added.

As such, she sees the threat landscape shifting in four areas. AI is a force multiplier, identity is the primary compromise vector, attackers are burrowing into foundational libraries and software supply chains “baked into the fabric of our digital world,” and nation-sponsored threat groups are learning more about points of weakness in enterprise systems, DeGrippo said.

Nobody is fully prepared for what’s coming and it would be naive or a bad defender mindset to think otherwise, she said. “This is a transformative period, and how organizations navigate that transformation is going to be make-or-break for a lot of them.”

The post Unit 42 warns AI has shifted balance of power from defenders to attackers appeared first on CyberScoop.

OpenAI says the behavior that led its agents to breach Hugging Face emerged in its research environment more than two months before the incident, and concluded that it was a failure of alignment as much as it was a failure of security.

The details come from a technical report the frontier AI company released Wednesday, which gives a full breakdown on how the incident unfolded and what the company has changed in response.

“This incident is the first known case of an automated agent collective acting offensively without authorization, and the autonomous cyber capabilities demonstrated represent a critical shift in the security landscape,” the report reads. “The central threat-model implication is that organizations should no longer assume that sophisticated cyber operations require continuous human direction, proceed linearly, or are constrained by the attention and coordination limits of individual human attackers. Agentic systems can persist across tasks, share discoveries, build on one another’s progress, and combine vulnerabilities, credentials, and permissions into attack paths that may not be apparent when individual weaknesses are assessed separately.”

According to the report, the earliest relevant activity occurred May 8, when a model in a training run was given a spreadsheet task that referred to files stored on a Google Drive. The agent had no internet access, which made the task impossible as written. OpenAI said it did not intend to create such a task. The agent continued reasoning and tried to use JFrog Artifactory, an internal package service and one of the few tools available to it, to retrieve the files. The attempt failed, but it wrote a file to the service.

On May 12, another agent in a testing environment could not reach a database file used to store information about protein sequences. The report says the agent reasoned that an agent in a different environment might have the file, and wrote a note into Artifactory reading, “Agent seeks [filename]; upload if found!” On May 13, other agents examining Artifactory began finding similar notes of that kind and adding their own, creating what OpenAI refers to as a message board.

By May 26, agents had used a flaw in Artifactory to reach the public internet, relying in part on information exchanged on the board. An internal team observed message board activity and disallowed internet access that month.

According to an independent report also published Wednesday by AI research nonprofit METR, Roughly 1200 agents communicated with one another on that message board, sending over 70,000 messages and files during the investigation period. Of those agents, 700 went on to participate in the attack on Hugging Face.

The Artifactory service itself was compromised in late June, when agents using AI exploitation benchmark ExploitGym identified a flaw in a legacy token refresh endpoint. From there, an agent traded a forged administrator token for a properly signed one, and by July 4 agents had a permanent account and scripts that ran commands as the service. A monitoring tool flagged the activity on June 27 but the run was allowed to continue, and only after the service failed on July 4 did OpenAI take it offline, revoke credentials and disclose the flaw to JFrog, resuming evaluations July 7.

From there, OpenAI’s models executed the attack on Hugging Face, which resulted in agents poisoning a dataset to run code on a processing worker, eventually gaining node-level access, and stealing cloud credentials.

The company says the Hugging Face incident was carried out by two models, including GPT 5.6-Sol, but was primarily driven by an internal model that is “trained to be highly persistent and diligent in its work.“

The internal-only research model is comparable in scale to GPT-5.6 Sol and was trained to advance persistence and multiagent collaboration, among other capabilities,” the report reads. “The model was not intended for public use, and was only used by a small number of OpenAI personnel for internal research.”

Lessons learned

As a result of the incident, OpenAI will now restrict network access in its testing, isolate risky projects, and increase monitoring of its reasoning systems. Staff will be alerted within 30 minutes if problems are detected, and unresolved issues will automatically pause the work.

The company will also use its own safety benchmarks on models it doesn’t publicly release. After the incident, tests with non-public models using ChatGPT’s consumer settings reduced the tendency to compromise infrastructure.

“We found the propensity to compromise infrastructure can drop over 100x when using the production ChatGPT harness and system prompt,” the report reads. “We also ran the auto-review system on a set of rollouts involved in the Hugging Face incident and found that it would have flagged a multitude of the models’ dangerous actions.”

The report comes as 15 state attorneys general wrote a letter earlier this month calling for OpenAI to release more details about the incident in order to “ensure the safety of its products poses an imminent risk of substantial harm.” Alabama Attorney General Steve Marshall went a step further Monday, issuing a subpoena to the company for more information.

The company wrote in the report that it hopes its findings will lead to industry-wide changes, especially as model capabilities continue to accelerate.

“The lessons from this incident extend to the entire AI industry,” the report reads. “As frontier models become more capable, the safeguards used to contain and monitor them must evolve as well.”

The post OpenAI: Agent behavior that led to Hugging Face intrusion formed in May appeared first on CyberScoop.

A cybercrime group responsible for a string of recent attacks against private equity firms, law firms and financial rating agencies remains active and continued to target new victims as of late last week, according to researchers.

BlackFile, which Google Threat Intelligence Group tracks as UNC6671 and associates more broadly with The Com, has been active since the start of the year, shifting its focus from one sector to the next. 

“We have seen continued targeting against the financial sector with additional targeting of other organizations including in the med tech space,” Austin Larsen, principal threat analyst at GTIG, told CyberScoop.

The extortion group impersonates IT support in voice-phishing and social engineering attacks, and recently split its extortion operations across four brands with shared infrastructure: Redact, Pink, Helix and Falcon.

Several organizations received new extortion demands from Redact in the last week, according to Google. 

BlackFile and its various affiliates have impacted organizations in multiple industries, including healthcare, technology, transportation, logistics, wholesale, retail and hospitality.

“BlackFile does go after some of the largest organizations in the sectors that they go for. They’re not going after small companies,” Larsen said. “This is big-game hunting.”

The group’s extortion demands often start around $3 million and payments, including several in the past few weeks, have typically been negotiated down to less than $1 million, according to Google.

Flashpoint researchers told CyberScoop they have observed malicious infrastructure targeting Blackstone, Bain Capital, Moody’s, CME and Apollo, but it’s unclear if any of those firms were compromised. 

BlackFile’s steady pace of activity underscores the persistent threat it poses, as it targets an average of 1.5 new victims daily, researchers said.

Some of the group’s recent victims have been subject to threatening messages and other forms of escalation, including swatting incidents, a tactic adopted by several subsets of The Com, according to Google. 

The attackers use hundreds of callers, often lower-level people that are recruited for a small fee or an opportunity to earn goodwill with the group, who make the voice phishing calls to obtain initial access. Larsen estimates less than a dozen core operators run the different brands under the BlackFile umbrella.

“From the intrusion data that we’re seeing, this does appear to be essentially the same group,” he said, adding that different people may be operating the various brands, but they’re all linked back to the same threat cluster using shared infrastructure.

Mandiant incident responders encounter BlackFile often, having been engaged by more than two dozen organizations successfully compromised by the threat group since January. New victims in the financial sector were calling Mandiant in for help earlier this month.

Voice-based phishing attacks for data theft extortion aren’t sophisticated or novel, but BlackFile and other cybercrime groups consistently prove their continued effectiveness across virtually any sector or organization. “They’re really hitting on the human weakness element here,” Larsen said.

The post Details emerge on BlackFile’s recent attacks on financial companies appeared first on CyberScoop.

As the White House and federal agencies grapple with frontier AI models and their hacking capabilities, researchers are warning that the industry’s “middle class” of smaller models may end up posing a greater threat over the long term.

Research from XBOW this week shows that a growing class of both proprietary and open-source models are becoming strategically important in the offensive security ecosystem. Models like Z.ai’s  open-weight GLM-5.2, xAI’s Grok 4.5, Anthropic’s Opus 4.7, Meta’s Muse Spark 1.1, still perform very strongly at many hacking and exploitation tasks that worry policymakers.

“It’s not even that the open-source variants or…not quite frontline competitors are catching up [to frontier models] as such,” said Albert Ziegler, head of AI at XBOW. It’s that they are crossing a certain threshold, which means that suddenly they are providing net value at a cheaper price.”

That wasn’t necessarily the case as recently as six months ago, when testing on mid-tier class models showed they struggled to complete “moderately complex” agentic tasks. Today’s middle class largely can. Their relative cheapness means users can spend many times more resources—running them repeatedly—to solve the same challenges. 

“Because these models are cheaper, it’s okay to give them more time, and they come from behind and leapfrog the big frontier model,” said Ziegler. “Now, that didn’t work half a year ago because…if you wanted to run some open-source model on a complex task in an agentic way…on a long horizon, then it would just get lost.”

GPT 5.5, now considered a near-frontier model, delivered one of the best performances on exploitation benchmarks that XBOW has recorded to date.

The jump between OpenAI’s GPT 5 and 5.5 “represented one of the clearest 2026 leaps in autonomous web application testing,” according to the report. It saw marked improvements over previous middle-class models in exploiting both “white box” and “black box” scenarios, or with and without access to the underlying victim source code. It also missed fewer vulnerabilities, with a “miss rate,” or failure to spot a vulnerability, of 10%, while GPT 5’s rate was four times larger, 40%.

The emergence of GPT 5.5 changed “the practical baseline for what frontier models can do in offensive workflows,” the XBOW report said.

But the performance leap goes deeper than that. GPT 5.5 performed higher in tests without source code access, while GPT 5 heavily leaned on source code. 

“That last result is significant: working without the code, as an attacker would, GPT-5.5 beat a prior version that could read it,” the XBOW report said. “What translated into findings was the ability to reach and prove a vulnerability against the running system, not to infer it from a pattern in the source.”

XBOW’s testing found that source code access was not as important to these models’ success as other factors, like live interaction with the actual website or software being exploited.

Frontier models like Mythos and GPT 5.6 are indeed more capable on individual cybersecurity tasks, but they can also come with exponentially higher token costs.

New research this week from Anthropic tested two models – Mythos Preview, which is used in Project Glasswing, and Opus 4.8 – to learn how quickly multi-agent swarms could find vulnerabilities in 15 open-source software projects when they coordinate and share information.

While a team of agents working individually and assigned to core directories found 21 vulnerabilities, the coordinating agent swarm found 266. But both tests had to burn through millions of tokens – 6.5 million and 27 million – to get there. Beyond the difficulties with getting access to frontier models, few individuals or organizations have the budget to underwrite that kind of research.

The way these systems coordinate can differ significantly from how humans work together.

Another experiment tested agents’ ability to coordinate on the development of a fantasy-themed video game. Earlier models, models like Opus 4.6, failed to properly coordinate and produced “bad” results, while later models like Mythos and Opus 4.8 were able to achieve better results but did so by hardly coordinating at all on tasks.

“The lack of coordination shown by agents in the fantasy game…in which they siloed themselves and largely failed to merge their work—roughly mirrors some ways in which humans can fail to coordinate,” the Anthropic blog stated.

Further, agents are more homogeneous than humans and “often act the same in situations where different people might take a much more diverse range of actions.”

XBOW also tested Mythos Preview, finding that it showed “exceptional” source-code reasoning and reverse engineering abilities, particularly with source code access. Like other models, losing live-site access had a big impact on its performance, and while Mythos is excellent at finding vulnerabilities, it’s less effective at exploiting them.

]Ziegler said the recent incidents at companies like OpenAI, Anthropic, Meta and others where frontier models escaped sandboxes and hacked into project-adjacent parts of the internet should rightfully alarm lawmakers, and demonstrate  the upper-tier capabilities of large language models.

Like most industries, cybersecurity favors cheap, high-performing tools over expensive ones. The widely adopted tools that have the most impact tend to be affordable and effective, not luxury products. 

And while these models still require human management to be wielded responsibly by law-abiding organizations, that cost tradeoff can look more attractive to malicious hackers, who tend not to care about collateral damage caused by their agents.

“Purely from an attacker’s perspective, I think we already are [there],” Ziegler said.

The post AI’s ‘middle class’ has gotten dramatically better at hacking appeared first on CyberScoop.

OpenAI announced Monday  it was expanding access to its frontier models for defensive cybersecurity, detailing different defensive and red-teaming workflows and a new partner program with major cybersecurity product providers.

In a pair of blogs posted Monday, OpenAI said it was updating its Daybreak program  – which provides unreleased frontier models to private organizations and governments for defensive cybersecurity work – and introducing a new model variant.

Daybreak Blue, powered by OpenAI’s ChatGPT-5.6-Sol, would operate with lower cybersecurity safeguards compared to other commercially available models and is described as “a recommended starting point for most defenders” that supports tasks like vulnerability discovery, secure code review, malware analysis, incident response and patch validation. 

Daybreak Red, meant for more advanced red-teaming, would provide access to a new model, dubbed GPT-5.6-Cyber, that the company said is more purpose-trained for finding vulnerabilities and testing (or exploiting) them. The model is also less likely to refuse requests around “dual-use cyber tasks.”

According to OpenAI, the organizations in Daybreak Red will have their use closely monitored and supervised, as GPT-5.6-Cyber is significantly more capable in carrying out malicious cyber tasks than Sol. A security evaluation the company devised tested both models on complex requests, including exploit chain development, authentication bypass, privilege escalation and other hacking tasks. Sol succeeded in 1.5% of the requests, while Cyber completed 95%.

OpenAI said it plans to publish a more detailed system card for GPT-5.6-Cyber at a later date.

“Models running with reduced safeguards carry risks beyond standard model usage, whether from misuse or misalignment,” the company said in a blog. “Despite these risks, we believe that democratizing access to frontier intelligence for defenders is crucial to accelerating and automating cyber defense.”

Additionally, OpenAI announced a partnership program with 16 major cybersecurity providers, saying organizations could access their models through their existing security services. The partners include IBM, CrowdStrike, Accenture, Ernst & Young, KPMG, Palo Alto Networks, Cisco, Cloudflare, Sophos and others. 

“These partners bring deep security expertise and established relationships with organizations around the world,” OpenAI said in its blog. “By bringing our frontier cyber models into their services, we can help more defenders find serious vulnerabilities, validate which ones matter, and fix them faster.”

Companies like OpenAI, Anthropic and others are trying to rebalance their priorities after a string of AI-agent sandbox escapes have rattled policymakers and caused some cybersecurity experts to question if AI companies are doing enough to properly isolate the models from the internet during testing. Last week, OpenAI said it was intentionally slowing down development of its newer “Astra” model in order to develop better guardrails to restrain its behavior.

Cybersecurity and AI experts have told CyberScoop that while AI systems have greatly improved at finding and exploiting vulnerabilities in software code, they still require substantial human guidance and supporting infrastructure to operate as intended.

Additionally, some research has shown that without such guidance, even near-frontier models can struggle to fully patch a discovered vulnerability or avoid introducing new bugs with their fixes.

The post OpenAI says Daybreak will expand to offer specialized cyber services  appeared first on CyberScoop.

U.S. and South Korean cyber agencies warned Monday about a ransomware-as-a-service outfit, Gunra, that reportedly recruits ethical hackers and penetration testers and benefits from North Korean government-linked hackers’ tools to target government and critical infrastructure organizations.

Gunra has gone after sectors such as academia, financial services and insurance, government services and facilities, healthcare, manufacturing and construction, media, retail, transportation and utilities. Its global scope is far-ranging, according to Monday’s alert: Africa, the Americas, the Asia-Pacific, Europe and the Middle East.

“Gunra is another variant in the ongoing trend of ransomware attacks causing disruption and harm to U.S. and international organizations,” said Chris Butera, acting assistant director for cybersecurity at the Cybersecurity and Infrastructure Security Agency, which produced the advisory with the Department of Defense’s Cyber Crime Center, FBI, National Security Agency, Secret Service and Republic of Korea’s National Police Agency.

The alert is part of the #StopRansomware series, a joint FBI-CISA project aimed at network defenders.

The FBI first took notice of Gunra in April of last year. The double-extortion group established a data leak site on Tor to list victims and publish purloined data. By January of this year, Gunra had launched a formal ransomware-as-a-service affiliate and was growing in its ambition, Monday’s alert states.

“The FBI observed the group adopting new branding aliases (notably operating under the name Golden Community) to support this expansion,” it reads. “Gunra has further commercialized its platform by actively recruiting penetration testers and ethical hackers to serve as initial access brokers, offering a share of the ransom profits in exchange for enterprise network access.”

Gunra seeks initial access with known vulnerabilities in internet-facing devices like firewalls or virtual private networks, and is based on or influenced by the Conti ransomware code leaked in 2022, according to the agencies.

Research published in July by a South Korean cybersecurity firm took note of Gunra overlap with Lazarus Group, although it doesn’t explicitly mention the latter group’s name.

“These commonalities suggest that although the state-sponsored threat group and the Gunra ransomware group appear to be separate threat actors with different ultimate objectives, they may have shared certain techniques, tools, and infrastructure or collaborated to a limited extent during the attacks,” AhnLab wrote in its report.

That kind of North Korean government-ransomware gang collaboration dates back to at least 2024. Nor is Gunra alone among ransomware-as-a-service outfits recruiting penetration testers.

The post U.S., South Korean government agencies caution to be on lookout for Gunra ransomware gang appeared first on CyberScoop.

As AI-generated code continues to be injected into all corners of the internet, concerns have risen about an expanding attack surface for malicious hackers to exploit.

Some have argued that the enhanced cybersecurity capabilities of large language models could serve as a check, finding and fixing vulnerabilities nearly as fast as they’re created.

But new research that tested the patching capabilities of two popular commercial models, OpenAI’s ChatGPT 5.5 and Anthropic’s Claude Opus 4.8, found that generative AI is more likely to create an exploitable patch or introduce entirely new bugs than close off a vulnerability.

Researchers at 1Password tested the models ability to patch six “high-impact, high-complexity” CVEs, including the “Copy Fail” vulnerability, a kernel flaw that can give an attacker root access to Linux cloud environments. The overall success rate (or fully patching the vulnerability without introducing new problems), was less than a coin flip at 47%.

“Our research findings show that, in aggregate across a variety of scenarios, both Claude and ChatGPT had a low rate of successful patch generation, which we define as full remediation of all known exploit paths with no erroneous changes to application behavior,” wrote Keith Hoodlet, Axel Mierczuk and Spencer Michaels.

“The models often addressed only a subset of vulnerable code paths, added fragile guard code that satisfied tests while failing to address the vulnerability’s root cause, and sometimes introduced subtle changes in the application’s behavior while patching the immediate vulnerability,” the authors continued.

The research suggests that largely autonomous vulnerability-discovery and patching may not yet be effective in fixing the explosion of vulnerable code that is being created in the AI era.

Other private sector research has pointed to a similar problem. A report this year from Veracode found that while LLMs have made “enormous strides” in crafting workable code, “security is a different story.” Testing across a range of frontier models found the average security “pass rate” for AI generated code is around 56%. Newer models like GPT 5.5 push closer to 70%, while more than half sit between 50-53%.

Veracode tested 100 different models and while there was variability, in general a small number of models were showing progress on security patching while the rest have experienced “stagnation.” Similar to the 1Password research, in 44% of Veracode tests the models introduced a detectable OWASP Top 10 vulnerability into the codebase.

An important caveat: neither report tested newer models, like Anthropic’s Mythos or OpenAI’s GPT-5.6-Sol, that frontier companies tout as having significantly higher cybersecurity capabilities.

Those advanced models can identify and fix vulnerable code. Anthropic and OpenAI are distributing them to key industries through Project Glasswing and Daybreak before foreign or open-source alternatives can compete.

Tim Jarret, vice president of product at Veracode, told CyberScoop that AI tools are still subject to a range of limitations that can make them unreliable for cybersecurity patching without knowledgeable humans in the loop.

While some vulnerabilities – like SQL injections – can be easily patched through automation, other bugs like cross-site scripting, can be exploitable in several different ways and require either a human touch, additional context or both to fully close off. Additionally, models can slowly lose context from prior sessions over time, affecting their ability to complete tasks correctly and raising the possibility they’ll hallucinate to fill in the missing gaps.

“I think we would say, at this point, that Iits premature to treat those as anything other than another code change to the code base that needs to be reviewed and accepted by the team, as opposed to letting the agent merge the code freely,” said Jarrett.

However, he acknowledged that may not be possible in a world where AI agents are generating exponentially more code for human defenders to review. Some kind of automated code review will be necessary – preferably not by the same automation tool that produced the code. The ultimate goal is the same as it has always been in security: “trust but verify.”

“Ninety percent of the time, the human check might just be ‘did the cross check look good?’ Do we have a thumbs up?’” Jarrett said. “In those cases where there’s still something wrong, that’s where you focus your attention a little bit more.”

The post More than half of AI-generated patches are broken appeared first on CyberScoop.

TeamPCP, the threat actor behind an unrelenting flurry of attacks on open-source software this year, has been active much longer than previously thought, according to research Oligo Security shared exclusively with CyberScoop. 

The threat actor, which gained notoriety and has captivated threat hunters as it compromised and injected malicious code into more than 1,000 software packages in less than four months earlier this year, was also responsible for attacks dating back to 2020, Oligo Security found. 

The security vendor’s research team found multiple attacks that bear the markings of TeamPCP, including a late 2025 campaign involving the exploitation of a ShadowRay vulnerability that resulted in the first self-propogating botnet running on hijacked AI infrastructure.

Evidence uncovered during that investigation into the ShadowRay 2.0 campaign was linked to more historical attacks originating from the same IPs, domains and other infrastructure TeamPCP used in attacks that captured widespread attention earlier this year. 

“The scariest thing in this campaign is the speed at which the payloads evolved and changed and adapted to the environment they run in. We saw changes in the speed that we’re not used to seeing in these kinds of attacks. They’re usually slow, careful,” said Uri Katz, director of research at Oligo Security. “This was clearly with the help of AI — the payloads changed rapidly to adjust and change to the environment that they were trying to attack.”

One of the domains that Oligo Security identified in July 2025 was in the profile of TeamPCP’s official GitHub account, said Avi Lumelsky, AI security researcher at Oligo Security. “It’s public, they’re not even trying to hide their identity,” he said. 

From there, Oligo linked TeamPCP to activity tracked under multiple names, including TA-NATALSTATUS and IronErn, spanning from 2020 to late 2025. Much of that activity was traced to the same IPs, domain names, a file server and command-and-control server, researchers said. 

TeamPCP emerged publicly as a brand in late 2025. Soon after, “TeamPCP started to go really broad and do campaigns, which are much more noisy,” said Gal Elbaz, co-founder and CTO at Oligo Security. 

Widespread adoption of AI and TeamPCP’s use of the technology supported this growth as the threat actor built a brand, got more active on social media and boasted publicly about its activities and claimed victims.

“The ability to control the infrastructure and orchestrate the attack with AI was also super new, and I’m sure it helps them,” Elbaz said. 

“All of the companies in the world are in this race to adopt AI because they are afraid their business will die, and they understand, of course, the opportunity. But it’s also what gives the attacker this power to go into it,” he added. “If you don’t really have visibility in what’s going on there or how it behaves, that’s exactly what attackers are after.”

TeamPCP’s more recent attacks have capitalized on new security gaps created by developers’ increasing reliance on AI and the automated systems companies use to deploy code. The threat actor is also consistently wrecking the open-source frameworks and software packages these systems rely on. 

“Most AI infrastructure is open source by design because nobody has the manpower and money to develop everything from scratch,” Lumelsky said. 

“We love open source. We use many of these products ourselves, but it’s all about reading the documentation, and I think many of these tools place the responsibility of using it right and security on the user, and developers are not used to these new kinds of animals,” he added. “That’s why the trust can be exploited at scale.”

As it uncovered a long operational history spanning multiple campaigns, Oligo Security has gained more confidence in understanding how TeamPCP operates. It also means TeamPCP was likely involved in other attacks that haven’t been attributed to it yet or attacks that haven’t been detected. 

“There’s a lot more out there that we haven’t caught or been able to prove up until now,” Elbaz said.

The post Open-source software’s archenemy TeamPCP goes back further than anyone thought appeared first on CyberScoop.

Like seemingly everything else these days, artificial intelligence will re-shape the way voters gather information on candidates running in the 2026 midterm elections.

In some ways, this is already the reality. Voters are increasingly turning to AI chatbots for information instead of Google.  Political campaigns are deploying deepfakes of their opponents. And AI systems have been developed to carry out increasingly complex  hacks.

Since the last major U.S. election in 2024, major tech companies have  embedded AI into their products while hundreds of millions of people have adopted the tools, either by purchasing subscriptions to commercial models or using open-source models. Yet both research and experts state that while AI systems have gotten better at handling basic facts, they’re nowhere near reliable enough to be a main source of  accurate or complete information. 

While chatbots are becoming a primary way that voters gather information on  local races, candidates, issues, and voting information, they are not substitutes for more authoritative sources, like a voter’s state or local election office. 

“I think this is one of the first elections we’re seeing…where AI is just everywhere,” said Thania Sanchez, senior vice president of research and analytics at the nonprofit States United Democracy Center. “Even if you just Google it, [now] the first thing that comes up is the AI overview.”

While AI companies have worked to cut down on errors in their model’s responses for questions around basic election information, they continue to fall short in important ways.

In new research shared exclusively with CyberScoop ahead of its release, States United Democracy Center tested two of the most popular tools — OpenAI’s ChatGPT’s free tier and the AI interface used alongside Google Search — for their performance on a series of basic questions around elections, such as how to register to vote, or a list of candidates in a race.

The models were chosen because they are free and easy to access. For Google AI, the nonprofit tested two types of accounts: ones running in Incognito Mode and ones that had a history of browsing election-skeptical websites.

The nonprofit ran two rounds of testing in 2025 and 2026, collecting nearly one thousand responses from the models submitted by users across six swing states (Arizona, Michigan, North Carolina, Nevada, Pennsylvania and Wisconsin).

In 2025 tests, 6.9% of responses from Google AI and 8.2% responses from ChatGPT“contained verifiable factual errors,” like not listing the correct candidates in a race or false guidance around polling site locations.

However, follow up tests in 2026 across Arizona, Pennsylvania and Michigan found that the error rates in both models had dropped to zero. The study notes that “this is real progress and should be acknowledged.”

But underneath those topline numbers, a more murky picture emerges around the tools’  reliability.

An AI response can sound accurate without actually being complete.  To wit: ChatGPT provided incomplete lists of current gubernatorial primary race candidates 88.9% of the time when queried.

Linking to a state election website – an output the study considers the single most important measure of voter utility  — happened less than 40% of the time. Whether due to formatting issues or the model ingesting outdated information, it’s a problem if voters use them as their primary information source for elections.

“It will be like ‘this person is the Republican candidate and this person is the Democratic candidate’ but it is not telling you there’s also these other third-party candidates,” said Sanchez. “It’s not giving you complete information, so the voter thinks these are the [only] two people running.”

A June survey from the Pew Research Center found that about half of U.S. adults reported having used chatbots at least once, up from a third in 2024, while a quarter reported using them daily. The top use case listed for engaging with the chatbot was searching for information.

Isabel Linzer, an elections policy analyst at the Center for Democracy and Technology, told CyberScoop that voters, campaigns and governments alike are using AI more freely and with fewer restrictions.

Bad actors in the information space have followed suit, and “we are in a phase now of generative engine optimization” where information operations are structured to rank higher in AI model responses.

“We’ve moved beyond [SEO] to [Generative Engine Optimization], and that’s where we’re seeing campaigns thinking about how to structure their materials to make sure that they are in a format that AI models want to use when they’re searching the web…to develop their responses to user queries,” she said.

There is also the underlying problem of frontier AI companies constantly tinkering with their models, their algorithms and the technologies they are intertwined with. . Election officials, by contrast, have decades of experience educating voters about their options.

A prime example of this churn occurred this past February, in between the first and second round of the study, when Google AI suddenly shifted to providing only links for election related queries in incognito mode, replacing the written summaries that showed up in the first round.

Like the study’s authors, Linzer said most people are still best served by going directly to local sources for accurate information on elections. With issues like ideological bias, the potential for bespoke or sycophantic answers for each user based on their prior chat histories and lack of predictability, voters should still be very careful about using AI chatbots as political truth machines.

The best thing that tech companies can do to educate voters is “making sure that for high-stakes situations like elections, that chats are connecting directly to the most important sources, like the website where you can actually register to vote,” said Linzer.

The post AI is getting better at election facts, but voters shouldn’t rely on it appeared first on CyberScoop.

In less than four hours early Tuesday, an attacker compromised a GitHub maintainer account and unleashed a self-replicating piece of malware which injected malicious code into more than 440 distinct npm packages, according to multiple security firms. 

The worm, built on the open-source Mini Shai-Hulud repository that TeamPCP published in May, was initially let loose in keyv, a data management interface software package with more than 600 million monthly downloads. The attacker spent the next 30 minutes compromising additional packages controlled by the same maintainer, including cacheable, flat-cache, file-entry-cache.

The attack spread to other maintainers, eventually compromising more than 860 packages with a “combined total of over 2 billion monthly installs,” Ilyas Makari, malware researcher at Aikido Security, wrote in a blog post. 

Wiz researchers told CyberScoop it hasn’t observed any new malicious packages since the initial wave moved through a massive footpoint of cloud and code environments in those first four hours. 

“This is the most critical initial compromise, with over 155 million weekly downloads on the root packages,” Wiz Research said in an email. 

Some of the compromised packages, including keyv, flat-cache and file-entry-cache, are present in more than 46% of all cloud environments, according to Wiz. “By comparison, back in the Shai-Hulud 2.0 campaign the most prevalent packages were only in about 28% of environments,” the company said. 

“Time will tell whether the eventual cost and impact outpaces past attacks, or whether adoption of hardening mechanisms such as package aging, and the usage of the relatively less aggressive Mini Shai-Hulud code as basis, will defray the final toll here,” Wiz Research added. 

Researchers from multiple firms sprung into action to monitor the widening attack spree and published indicators of compromise to help potential victims hunt for malicious activity in their systems. 

The Mini Shai-Hulud variant used in these attacks scoops up a trove of sensitive data, including npm, GitHub, AWS and continuous integration credentials. It also steals AI-related configuration files and cryptocurrency wallets, researchers said. 

Microsoft, Aikido, Socket and Wiz all said the same payload and pattern was observed across all affected packages, indicating a single attacker or threat cluster was behind the supply-chain attack and using multiple stolen tokens. 

The malware showcased a few pieces of new functionality, but retained the same core mechanisms that are hallmarks of Mini Shai-Hulud. 

“The evolution is consistent with what we’ve seen from them in past waves, however we don’t yet have the hard links” to confidently attribute the attacks to TeamPCP, Wiz Research said.

The notorious threat actor, which Google previously told CyberScoop it attributes to one core operator that was located in South Africa during at least some of the attacks, compromised and injected malicious code into more than 1,000 software packages in less than four months earlier this year.

The post Massive supply-chain attack compromises 440 packages under four hours appeared first on CyberScoop.