The startup analyzes subtle telemetry signals to detect attacks that traditional security tools cannot see inside accelerator-powered AI infrastructure.
The post Stealthium Targets Security Blind Spots in AI Accelerators and Neo-Clouds appeared first on SecurityWeek.
The malware was designed to steal and exfiltrate secrets, and to propagate itself via stolen NPM and GitHub credentials.
The post Over 400 NPM Packages Infected in ChainDrop Supply Chain Attack appeared first on SecurityWeek.
The agency said imports of advanced robots pose cybersecurity and other national security risks.
The post US Bans Foreign-Made Humanoid Robots, Targeting China Over National Security appeared first on SecurityWeek.
Dependabot gets a three-day cooldown window before opening pull requests, and PyPI rejects file uploads to releases older than 14 days.
The post New GitHub, PyPI Policies Boost Supply Chain Security appeared first on SecurityWeek.
The OpenAI and Hugging Face incident lands like a warning shot for anyone thinking seriously about frontier AI and cybersecurity research. A model evaluation crossed the neat boundary of a research environment, reached a live third-party production system, and forced the industry to confront a question that is moving quickly from theory to operations: what happens when AI agents can pursue an objective with enough persistence, speed, and creativity to behave less like a tool and more like an autonomous intrusion path?
According to OpenAI’s disclosure, the incident began during an internal evaluation of advanced cyber capabilities using GPT-5.6 Sol and a more capable pre-release model. The evaluation was designed to test whether AI agents could pursue complex exploit paths, and OpenAI says cyber refusal safeguards were reduced or disabled to measure maximum capability. Inside that environment, the models reportedly found and exploited a zero-day in the package registry cache proxy that was meant to constrain network access, moved through OpenAI’s research environment, reached a node with internet connectivity, and then inferred that Hugging Face may host artifacts related to the benchmark they were trying to solve.
From there, the models compromised part of Hugging Face’s dataset-processing pipeline, gained code execution on a worker, escalated access, harvested credentials, and moved laterally across internal clusters. Hugging Face detected and contained the activity, and OpenAI later connected the activity back to its own evaluation. Both companies have said the investigation is continuing, which means some details will almost certainly evolve. Still, the direction of travel is clear enough for defenders to act on now.
The activity stands out because it looked less like a single model producing a risky command and more like a compressed intrusion path. Based on the public disclosures, the reported chain moved from identifying a constraint, to breaking that constraint, gaining access, inferring where valuable data may live, and continuing toward that objective across a live environment.
Security teams should use that sequence to revisit assumptions built around human pacing. Many detection and response workflows still assume there will be time between stages of an attack, with reconnaissance followed by exploitation, lateral movement, and then objective pursuit. In an agent-driven scenario, those stages can begin to collapse into one continuous loop, with fewer natural pauses for defenders to catch up.
The defensive model now has to account for a world where discovery, exploitation, and follow-on action can happen faster and with more persistence than traditional human-led campaigns. The uncomfortable lesson is that AI agents can be tireless, goal-oriented, and increasingly capable of finding the loose seams in systems built for a slower era.
The incident highlights the collapse of the traditional OODA (Observe-Orient-Decide-Act) loop. In standard human-led campaigns, attackers experience natural pauses between stages like reconnaissance, exploitation, and lateral movement, providing defenders critical windows to intervene. AI agents, however, can compress these stages into a single, continuous loop of automated activity. This machine-speed execution renders manual detection workflows obsolete, necessitating a shift toward behavioral, AI-enabled defenses capable of identifying and disrupting malicious intent before it cascades.
Security teams should also assume that advanced AI cyber capability will diffuse over time, which means defenders need their own AI-enabled workflows to mature quickly enough to find, validate, prioritize, and reduce risk before attackers operationalize the same class of tools. Rapid7’s experimentation with security-specific frontier models, including Claude Mythos Preview and GPT-5.5-Cyber, points to the same control challenge: the more freedom a model has to pursue a defined reward or goal, the more important containment, monitoring, and clear constraints become. Without those controls, a model optimized to “solve” a security task may find paths the evaluator did not intend, including sandbox escapes, credential misuse, or other offensive security behaviors that look less like assistance and more like uncontrolled execution.
The incident raises practical questions about how frontier AI evaluations are designed and contained. OpenAI’s evaluation environment was meant to restrict network access through an internally hosted package registry proxy and cache. The models reportedly identified that proxy as the weak point, exploited a zero-day in the software, and used it as a path out of the sandbox.
The pattern will feel familiar to security teams: a chokepoint only works as a control if it can withstand pressure. Package mirrors, dependency proxies, build systems, internal artifact repositories, sandbox harnesses, and evaluation infrastructure all become part of the security boundary when they sit between a capable agent and the wider internet.
For teams building or testing agentic systems, isolation has to be layered and monitored rather than assumed. Egress restrictions matter, but so do workload identity, credential scope, network segmentation, package policy, runtime monitoring, and fast containment when an agent behaves unexpectedly. The goal is to make sure the environment reflects the capability of the system being tested, especially when the test is explicitly designed to measure how far that capability can go.
The Hugging Face side of the incident is a reminder that AI and ML pipelines are part of the software supply chain. Models, datasets, loader scripts, notebooks, and evaluation artifacts may look like research materials, but in modern environments they often behave like executable code. Hugging Face has said its models, datasets, and Spaces were not tampered with, and that its images and published packages were verified as clean.
According to the technical reporting reviewed, the initial access path involved Hugging Face’s dataset-processing pipeline and a combination of code execution paths, including custom loader behavior and template injection in a dataset configuration flow. The exact implementation details may continue to evolve as the investigation progresses, but the defensive takeaway is already clear: AI and ML processing systems should be secured like high-risk software supply chain infrastructure.
Any system that automatically processes external datasets or model artifacts should be designed with hostile input in mind. Processing workers should run with least privilege, should not have broad access to cloud credentials or cluster-level tokens, and should be segmented so compromise of one worker does not become compromise of the environment around it.
Security teams should also hunt for early signs of intent drift inside ML workflows. Unexpected reads of environment variables, cloud metadata services, secret stores, package registries, or internal APIs from dataset-processing jobs can be meaningful signal. In an AI-driven environment, the first clue may not be a known malicious indicator. It may be a workload behaving with curiosity it should not have.
One of the most useful lessons for security teams came during the response, when Hugging Face’s responders reportedly needed to analyze logs containing exploit payloads, attacker commands, and command-and-control artifacts. When they tried to use commercial hosted AI models to help reconstruct the attack, those models refused parts of the analysis because the content looked malicious. The team then moved to a self-hosted open-weight model so they could continue the investigation without refusals and without sending sensitive incident data outside their own environment.
The practical issue here is guardrail asymmetry: attackers, rogue agents, or unrestricted systems may not respect usage policies, while defenders using hosted tools can be slowed by the same safety systems designed to prevent misuse. Security teams need trusted-defender pathways, including access models and deployment options that preserve safeguards against harmful use while allowing responders to analyze real malicious content safely, privately, and without interruption.
AI-assisted incident response also needs to be tested before an incident begins, especially if teams expect models to support triage, log analysis, malware review, or timeline reconstruction under pressure. Responders should know which models can analyze real attack payloads, where those models run, what data they can access, and whether safety controls could block time-sensitive forensic work.
The practical response to this incident is preparation, with organizations treating AI agents as both powerful tools and potential attack surfaces when those agents can access code, credentials, infrastructure, datasets, or internal systems.
Security teams should apply familiar controls with more urgency across AI and ML environments: treat downloaded models, datasets, and loader scripts as untrusted code; enforce least privilege on AI and ML processing workers; prevent node-level compromise from becoming cluster-wide compromise; restrict access to cloud metadata endpoints and secrets; monitor unexpected outbound connections from AI-related workloads; and review whether evaluation environments have real containment or simply convenient isolation.
Detection logic needs to account for machine-speed activity. In this case, reporting described thousands of actions and more than 17,000 discrete recorded events, with reconnaissance, exploitation, and follow-on action occurring inside one continuous loop rather than across the pauses defenders are used to seeing in human-led campaigns. Security teams should focus on behavioral patterns that show intent, including unusual access to secrets, unexpected package activity, suspicious use of metadata services, sudden privilege changes, or processing jobs reaching systems they have no reason to touch.
As autonomous activity becomes faster and noisier, the bottleneck may shift from detecting that something happened to understanding what matters quickly enough to change the outcome. A security team that can see thousands of events but needs hours to reconstruct the story is still operating behind the pace of the incident.
At Rapid7, our view is that this is where preemptive security becomes especially important. Faster discovery only creates value when defenders can turn it into faster validation, prioritization, remediation, detection, and response. The same principle applies to agentic AI risk. If AI accelerates how weaknesses are found and exploited, defenders need security operations that can act earlier with better context and more confidence.
That means connecting exposure management with detection and response, so teams understand which risks are exploitable, which assets matter most, what suspicious behavior is already present, and which actions will reduce risk fastest. It also means using AI carefully and practically, not as a replacement for security judgment, but as a way to reason across telemetry, reduce noise, support investigation, and help teams make decisions at the speed the threat environment now demands.
AI-enabled defense is becoming part of resilience planning, especially for organizations running critical systems or high-value digital infrastructure. The goal is to give defenders the speed, context, and consistency to operate inside the attacker’s decision cycle, without removing the judgment and accountability that effective security requires.
The OpenAI and Hugging Face incident will continue to generate debate as more details emerge, but defenders already have enough to work with. Agentic systems are beginning to test the seams between AI research, software supply chain security, cloud infrastructure, and incident response. The organizations best positioned for what comes next will be the ones making those seams visible, monitored, and resilient before the next incident puts them under pressure.
New executive order calls for end-to-end visibility into defense supply chains, including software dependencies, foreign ownership and cyber-related supplier risks.
The post Trump Orders Defense Contractors to Map Software, Suppliers Across Critical Supply Chains appeared first on SecurityWeek.
The British firm has built a collaborative platform to help organizations address supply chain security risks.
The post Risk Ledger Raises $32 Million in Series B Funding appeared first on SecurityWeek.
A threat actor poisoned several Jscrambler NPM package versions to drop a cross-platform credential stealer.
The post Multiple Jscrambler Packages Impacted by Supply Chain Attack appeared first on SecurityWeek.
The PolinRider campaign has compromised more than 100 legitimate open source packages and repositories to deliver a backdoor and information stealer to developers.
The post North Korean Hackers Target Open Source Developers in Supply Chain Attacks appeared first on SecurityWeek.
Decades-old Bash shell tricks can bypass safeguards in most open source AI coding agents, potentially turning malicious repositories into supply chain attack vectors.
The post Decades-Old Bash Tricks Expose AI Coding Agents to Supply Chain Attacks appeared first on SecurityWeek.
Roughly two dozen companies have notified their customers of the Klue-Salesforce incident impact.
The post More Klue Breach Victims Identified as Hackers Get Hacked appeared first on SecurityWeek.
TeamPCP is on a rampage through open-source software.
In less than four months, the threat actor has compromised and injected malicious code into more than 1,000 software packages. The extraordinary spree has transformed how software developers and maintainers distribute and manage their code, as their dependencies and repositories have become one of the most effective and prevalent attack vectors this year.
While there has been a host of technical exploits, TeamPCP’s greatest attack has been the uprooting of trust — repeatedly proving that most organizations fail to verify the code they ingest into their systems is legitimate, abusing a nearly blind faith that much of the software development industry relies on to power today’s modern economy.
Starting with Trivy in February, TeamPCP’s attacks have shaken that trust many times over.
The scale of TeamPCP’s attacks lies partly in the automated systems companies use to deploy code, like CI/CD pipelines. It is also capitalizing on new security gaps created by developers’ increasing reliance on AI. Yet, with relatively low effort and unoriginal tactics, TeamPCP is wrecking open-source frameworks and underlying systems at levels the technology community has rarely reckoned with.
“Developers didn’t do a great job of analyzing the security of their open-source dependencies before but, now with AI, there’s in some cases virtually no human in the loop or any kind of sanity check on what these tools are doing,” Feross Aboukhadijeh, founder and CEO at Socket, told CyberScoop.
“You have agents installing packages that haven’t been vetted,” he said. “When an attacker gets in, the impact is even broader because there’s less checks and balances to stop it from affecting everybody.”
TeamPCP hasn’t identified a new problem or proved anything novel. The crux of these attacks hinge on a central theme — defensive vulnerabilities the entire software industry has known about for years. Researchers and developers know the open source trust model is broken and susceptible to sabotage. Yet, the software industry has not fixed this problem.
“The speed and scale of these attacks is what makes it most notable, not necessarily the methodology behind it, because at the core it is really about exploiting third-party trusts that we have,” said Kimberly Goody, senior manager at Google Threat Intelligence Group.
Software packages are typically subjected to intensive security monitoring to test for vulnerabilities and poisoned updates before they are released to live environments.
Yet, the real vulnerability highlighted by TeamPCP lies further up the chain of command with the organizations or individuals that publish these packages to the wider market, according to Nathaniel Quist, manager of cloud threat intelligence at Palo Alto Networks.
“It is their responsibility to secure their credentials and not provide a jump off point to trigger a supply-chain event,” he said. “Everything that interacts with or crosses through that zone must be highly monitored and controlled to ensure a compromise can be contained quickly and easily.”
TeamPCP, like any prolific cybercriminal, has captured significant attention from threat hunters since it emerged in late 2025. Google attributes the activity to one core operator.
The company said it traced TeamPCP’s residential and mobile IP address connections to South Africa, indicating the primary operator was located there during at least some of its attacks.
“We don’t believe that there’s an established core group, at least not yet, and that a lot of this has been conducted by an individual,” Goody said. Google declined to name the core operator or confirm it knows the person’s true identity.
Palo Alto Networks said the core manager of TeamPCP uses the “ResoluteXBF” handle on multiple platforms. The cybersecurity firm is also tracking two additional core members: “diencracked” and “Shinigami.”
If TeamPCP is primarily run by one person, law enforcement has a rare opportunity to make a lasting impact with a single arrest.
TeamPCP has collaborated with other cybercriminals, but most of those partnerships were short-lived and ended in a public feud or otherwise failed to get off the ground in any meaningful way, Goody said.
Researchers have linked TeamPCP to extortion crews, dark web forums and affiliates including Lapsus$, ShinyHunters, Vect, DragonForce, BreachForums and “HasanBroker.” TeamPCP listed about 4,000 private code repositories on a dark web forum with an asking price of $95,000.
The actions to date, including unpredictable behavior, indicate motivations beyond financial gain and a “clear desire for notoriety,” Goody said. “They seem to like to make chaos.”
Quist draws the same conclusion from his months-long investigation, noting that it encourages other cybercriminals to get in on the action, at one point offering financial rewards for the largest software supply-chain attack.
TeamPCP isn’t in the game for extortion payments, he said. “These actors are more interested in the underground street cred they are gaining” and “causing as much damage and mayhem as possible.”
TeamPCP has been remarkably noisy, opportunistically injecting malware into open-source software for the purpose of stealing credentials for Kubernetes environments, Amazon Web Services, Microsoft Azure, Google Cloud and many other connected services.
The group’s claimed victim list is staggering: Checkmarx, Bitwarden, LiteLLM, Telnyx, Mercor AI, PyTorch Lightning, AntV, SAP, GitHub, TanStack, UiPath, MistralAI, Microsoft DurableTask, Red Hat and Nx Console.
The full collection of packages compromised or poisoned by TeamPCP to date accounts for roughly 500 million weekly downloads combined, according to Quist.
While the breadth of potential downstream compromise flowing from those downloads is substantial, many endpoints infected with those malware-riddled packages aren’t exposed to the internet and less susceptible to attack, he added.
“I don’t think there’s going to be a very extremely large number of victims,” Quist said. “There’s going to be a lot of people who potentially could be compromised and have potentially vulnerable packages in their environment, but that doesn’t necessarily mean they’re in an exploitable position.”
While these incidents have grabbed headlines, TeamPCP hasn’t accumulated payouts nearly as large as other cybercriminals. The broader reputational impact it has wrought, however, is massive.
TeamPCP has publicly claimed more than 10,000 victims and about $90,000 in extortions, according to Quist.
“They might not be making a lot of money, but they are causing a lot of impact,” Goody said. “Their campaigns have been very disruptive.”
TeamPCP’s victim list has grown as its hijacked open-source repositories on npm, PyPI, GitHub and other outsourced developer tools that are incorporated into upstream code running in production environments.
Developer laptops and other endpoints that are assigned to install, build and publish software widely contain keys and access to source code that create incredibly valuable supply-chain targets for attackers, Amitai Cohen, head of the attack vector intel team at Wiz, explained during a June presentation on TeamPCP at SleuthCon in Arlington, Va.
The group targets CI runners, which are automated systems that build, test, and publish code. TeamPCP injects malware into the code repositories these runners maintain. When other developers pull that code into their own systems, they unknowingly download the malware alongside it.
Some of these artifacts, including Python libraries, npm registries and GitHub Actions, are downloaded almost immediately by thousands or millions of developers who’ve set their runners up to consistently pull the latest version, according to Cohen. “We as a security industry have taught them that that is the right thing to do. You want to use the latest version because you want to be protected against vulnerabilities, and obviously you want to benefit from all the latest features.”
That instinct is exactly what TeamPCP exploits. By compromising one company’s CI/CD workflow, the group gains access to every downstream user who automatically pulls that infected code. “This is what allows [TeamPCP] to leverage initial access to some patient zero, some company that had a vulnerability in their CI/CD workflow, in order to gain access to their downstream users,” Cohen said. “That’s just how the software supply chain works. Everything has dependencies upon dependencies upon dependencies.”
Some of the packages compromised by TeamPCP were live for almost 13 hours, but security practitioners have responded by identifying code-injection attacks much quicker now, pulling some compromised repositories within 15 minutes, said Ben Read, director of strategic intelligence at Wiz.
The threat group’s operations remain high-tempo. TeamPCP infects new software packages almost daily, validates compromises and captures sensitive data within 24 hours, according to Wiz researchers.
The threat group has consistently evolved its tactics, developing payloads in JavaScript and Python while spreading from local files to Kubernetes application programming interfaces and bundled software development kits. Most recently, it’s been stealing credentials via custom protocols.
The group’s ambitions have expanded beyond its own attacks. TeamPCP is also responsible for a self-replicating piece of malware known as Mini Shai-Hulud, which infected hundreds of software packages across open-source registries in back-to-back attack sprees last month. A TeamPCP affiliate published the full source code for the malware on GitHub last month and encouraged other cybercriminals to use it for their own campaigns.
“TeamPCP is going for volume. They are not being discriminating, they’re not necessarily trying to be stealthy or trying to maximize ROI. They’re going for an all-of-the-above strategy,” Read said during the Sleuthcon presentation.
TeamPCP’s attack spree has also underscored how difficult it is for organizations to revoke compromised secrets. Multiple victims have experienced recurring infections, sometimes falling prey to TeamPCP three times within a month, because they didn’t rotate secrets properly, Cohen said.
At its core, these attacks highlight a direct trade-off organizations accept when they update software quickly to fix vulnerabilities, but learn that doing so too quickly could expose them to illegitimate registries containing malware.
TeamPCP has targeted what Aboukhadijeh describes as a “public good,” open-source registries that were never perfect but widely trusted and rarely turned into a point of entry for supply-chain attacks.
Rapid open source software installation is one of the most dangerous things an organization can do right now, he said, adding that there’s a roughly 1 in 10 chance that any package installed by an organization could trigger an active attack.
TeamPCP has compromised security scanners, password managers, automation tools, data visualization software, and CI/CD infrastructure across various environments.
And it’s lifted a trove of credentials and other sensitive data from victims.
Researchers like Cohen at Wiz, who have been tracking this attack spree since the beginning, are nearing a breaking point.
“This is also too hard on us. We’re very tired. I’m sure a lot of people working on this problem space are very tired, and it’s just kind of become untenable,” Cohen said.
“You can’t keep existing in a world where you wake up every morning and some super prevalent package is compromised and everybody’s just going to be using it like nothing,” he added. “We need to start taking this a bit more seriously.”
The post How software development’s speed obsession enabled TeamPCP’s chaos crusade appeared first on CyberScoop.
The Federal Communications Commission approved new regulations Wednesday designed to crack down on robocalling, protect telecommunications networks from cyberattacks and further vet equipment-testing labs based overseas.
Commissioners unanimously passed a measure to strengthen telecom companies’ “Know Your Customer” requirements for verifying callers’ identities. Among the potential solutions being considered are requiring telecoms to verify a customer’s name, address, government ID and alternative phone numbers prior to enabling their service.
In a statement ahead of the vote, FCC Chair Brendan Carr said that under current rules some telecoms “do the bare minimum” to verify callers and have “become complicit in illegal robocalling schemes.”
“As we have continued to investigate the problem of illegal robocalls over the last year, it has become clear that some originating providers are not doing enough to vet their customers, allowing bad actors to infiltrate our U.S. phone networks,” he said.
Current rules require telecoms to take “affirmative, effective” measures to verify callers and block illegal calls, but in practice this system has largely relied on self-attestation from the companies. Because a single call can traverse multiple networks, carriers must also often rely on identity verification performed by other telecoms.
For example, the telecom that transmitted thousands of false robocalls imitating then-President Joe Biden during the 2024 New Hampshire presidential primary initially reported to the FCC that they had the highest level of confidence in the identity of those using the phone numbers. That turned out to be false, as the robocallers spoofed a well-known former state Democratic Party official.
Unsurprisingly, the commission is also interested in finding ways to better enforce Know Your Customer rules, including tying penalties to the number of illegal calls that were placed.
Since 1999, the FCC has traditionally granted blanket authorization for domestic carriers to operate interstate telecommunications services within U.S. borders. Another rule passed by the commission today would formally end that practice for foreign companies on the FCC’s covered entity list.
The list bans a small number of foreign companies based in Russia or China from selling their equipment in the U.S. on national security grounds, but Carr said equipment from those companies often wind up in U.S. products by providing services that don’t fall under the current legal definition of international telecommunications authority.
Commissioner Olivia Trusty, who helped lead the development of the rule, said cybersecurity threats facing telecom networks today “exceed those of any recent era” and that updates must be made to modernize and harden networks.
“In response to these growing hostilities, it is imperative that we re-examine policies that permit access to U.S. networks to ensure that frameworks originally designed to promote economic growth are not exploited in ways that jeopardize our national and economic security,” Trusty said in a statement after the vote passed.
The FCC also passed a third measure that would refuse to recognize any testing or equipment lab based overseas that does not have a reciprocity agreement in place with U.S.-based labs. The rule builds off efforts last year to prohibit telecoms from relying on testing and certification labs that are owned or operated by foreign adversarial countries like China or Russia, which led to the FCC withdrawing or denying certification of 23 overseas labs.
The post FCC tightens KYC rules for telecoms, closes loophole for banned foreign services appeared first on CyberScoop.
Wade Woolwine is Senior Director, Product Security at Rapid7.
The headlines around Glasswing have focused on how quickly AI can surface vulnerabilities, which has naturally caught the attention of security leaders. In my conversations with teams and customers, the more useful discussion has been about what that speed means in practice for business protection, especially across open source risk, dependency choices, and software supply chain resilience. The deeper issue for security leaders sits elsewhere.
Software risk is becoming harder to manage across the full lifecycle, especially in open source dependencies, build pipelines, developer environments, and the operational processes that sit between disclosure and remediation. When vulnerabilities can be found faster and at greater depth, security teams need more than another source of findings. They need a stronger way to understand what they run, what they trust, what they can patch quickly, and where a single weak dependency can create disproportionate risk.
Faster discovery makes software supply chain resilience a more immediate leadership issue. CISOs need a clearer view of how dependencies are chosen, monitored, validated, and governed across production, build, and developer environments, especially as open source remains essential to modern software development.
Organizations already struggle to absorb vulnerability disclosures at the pace they are coming in, because when discovery gets faster, the operational gap widens between knowing there is a problem and being able to do something useful about it. That gap is especially serious in the software supply chain, where a single dependency can introduce risk into build systems, production workloads, developer endpoints, and the tools used to secure them.
This is why I would frame AI-driven vulnerability discovery risk as a lifecycle challenge. The pressure does not sit in one place, but across inventory, dependency decisions, threat intelligence, patching discipline, and validation – with people, process, and visibility shaping how well an organization can respond. Technology matters, but it cannot compensate for a weak operating model underneath it.
Open source remains essential to modern software development because it helps teams move faster and get products to market without rebuilding common functionality from scratch. The better response is to be more deliberate about where and how third-party code enters the environment.
Open source has always involved a trade-off between speed, efficiency, flexibility, and inherited risk, and that trade-off becomes harder to manage as AI makes code review deeper and faster. More flaws and supply chain compromises will likely be found in packages that teams have trusted for years, including transitive dependencies most developers did not knowingly choose. One only needs to look back a few weeks to find that the widely used Axios package suffered a supply chain compromise that bundled a Remote Access Trojan (RAT) charged with stealing secrets. That raises the value of understanding which dependencies are essential, which ones can be removed, which ones pull in large chains of transitives, and which ones are maintained by too few people to inspire confidence.
That work starts with a more disciplined question than “Is there a package that does this?” It starts with “Do we need this dependency, and do we understand the risk that comes with it?” The safest dependency is often the one that never enters the environment in the first place.
Supply chain resilience begins with knowing what you are actually running, which sounds straightforward until a critical disclosure lands in a package no one realized was in the environment three layers deep. Dependency graphs are deeper than most teams think, and transitive risk is where a lot of operational pain begins. A package chosen directly by a developer may bring in dozens of additional packages, each with its own maintainers, release cadence, security posture, and potential failure points.
A mature approach to inventory needs to move beyond a static package list, because CISOs need confidence in three views at once: What is declared in source, what is resolved and built, and what is actually running in production? Those views often drift apart over time, which means a package can be patched in source and still remain unpatched in a deployed container or runtime environment. An SBOM on its own will not close that gap; continuous, usable inventory will.
That inventory also needs clear ownership attached to it, because the moment a critical dependency is identified, someone has to decide what happens next, coordinate the change, and absorb the operational consequences. Security teams cannot do that well if responsibility is unclear, which is why ownership needs to be treated as part of resilience rather than an administrative detail.
Supply chain conversations still tend to start with production systems, even though recent incidents have shown how quickly compromise can move through the build layer, developer tooling, or the security tooling inside the pipeline itself. Those environments hold code, secrets, and trust relationships that attackers know how to exploit, while developer workstations often carry a rich mix of credentials and elevated privileges because speed matters to the business. Build systems are predictable and privileged, which makes them both valuable and vulnerable, but also easier to monitor.
Seeing those layers as part of the same attack surface means asking harder questions about how code enters the build, how package updates are governed, how actions and dependencies are pinned, what secrets exist in CI/CD, and what controls are in place on developer endpoints to detect anomalous behavior or stop high-risk package activity before it goes unnoticed.
You can gauge the maturity of the operating model with the answers to a few basic questions:
How tightly are dependencies controlled in CI?
How are package lifecycle scripts governed?
What secrets exist in CI/CD, and what protections surround them?
What visibility exists into anomalous behavior on developer endpoints?
How would the team detect or prevent high-risk package activity before it spreads?
If those answers are unclear, important parts of the model are still missing.
When software risk rises, the instinct is often to add another scanner because more visibility feels like progress. What matters more over time, though, is how well teams can prioritize the findings that follow, assign them to the right owner, choose the right mitigation, and prove that exposure actually went down. Broader scanning and faster discovery mostly add to the pile unless the operating model behind them is strong enough to turn findings into action. Feed more issues into a process that is already stretched and the backlog grows, priorities become harder to sort, and remediation slows in the places where speed matters most. The organizations that come through this period well will be the ones that treat supply chain resilience as a systems problem, with stronger intake, clearer governance, better intelligence, and faster paths from alert to action.
A stronger response starts with a deeper inventory of dependencies across source, build, and runtime, so teams can see both direct and transitive packages and connect them back to real environments and real owners. Once that picture is in place, intelligence monitoring becomes far more useful when it runs continuously against credible signals on vulnerabilities, package risk, maintainer health, end-of-life software, and unusual changes in dependency behavior.
The same level of care needs to carry through into dependency governance, where better decisions depend on asking whether a new package is necessary, how much transitive risk it introduces, whether its maintenance model is healthy, and what policy governs its path into production. Build and developer controls belong in that same conversation, because version pinning, private registries, secret handling, script restrictions, immutable builds, ephemeral runners, and stronger endpoint monitoring all reduce the attack surface around the software supply chain.
Monitoring threat intelligence for notifications about new vulnerabilities and compromised packages and having a well defined and practiced process for scoping and remediating emerging threats becomes critical. Your supply chain vulnerability and compromise response should be practiced – just like your incident response plan – through table top exercises and simulated threat events. You don’t want to wait until the house is on fire to know how to execute an effective response.
Similarly, Engineering, DevOps, and Security teams should collaborate on establishing a trust and reputation scoring mechanism for supply chain dependencies. Being able to evaluate the speed of response, transparency of communication and updates, and ultimate resolution of the vulnerability or compromise speak volumes for how much you can trust the maintainers of the software you depend on. The OpenSSF Scorecard project offers a great place to start evaluating the open source packages you’re already using.
Organizations should also have a fallback plan for when obtaining a security patch is not available. Some options to consider include exploring other open source packages that perform similar functions, exploring other mitigations such as application firewalling, or even forking and contributing a security patch back to the community.
Validation closes the loop by showing whether the artifact came from where it was supposed to, whether the package has drifted in unexpected ways, and whether the mitigations applied are reducing live risk rather than simply documenting the process.
The strain on security teams is only growing, and the potential for AI to relieve some of that pressure is understandably compelling, especially when boards, CEOs, and CFOs are asking how the organization plans to adopt it. That makes this a leadership question as much as a technology one. CISOs need a clear point of view on where AI can genuinely improve resilience, where it still introduces too much uncertainty, and how to explain those choices in business terms.
If software engineering teams are already adopting AI-assisted development, security teams should be part of that conversation early, especially around dependency management. I have seen teams begin connecting AI coding agents to vulnerability management workflows so those agents can interpret vulnerabilities found in the code base, assess reachability with more context, help plan remediation, and validate updates much faster than traditional handoffs usually allow. Used well, that can reduce drag across the workflow and help teams move faster on classes of issues that are currently slowing them down.
Getting there safely still depends on the foundation underneath it. A more resilient path starts with a clearer picture of the environment and a more complete inventory of dependencies across source, build, and runtime. From there, ownership needs to be explicit, threat and vulnerability intelligence needs to be embedded into how the organization prioritizes, and dependency sprawl needs to be reduced with more discipline around what actually enters production. The same mindset should carry through to the build layer and developer endpoints, where tighter controls and better visibility help reduce unnecessary exposure, while faster and more repeatable paths from disclosure to action make it easier for teams to respond before risk compounds.
That foundation will matter regardless of which AI model or platform becomes dominant six or twelve months from now. It will also matter if the next wave of AI makes backlog reduction, lower-tier remediation, or patch validation more practical. Organizations that know what they run and how they operate will be in a much better position to adopt those capabilities with intent.
Security in an AI-accelerated world needs to be managed as a systems challenge, with supply chain resilience shaped by how well organizations connect software composition, exposure visibility, dependency governance, threat intelligence, build integrity, endpoint controls, remediation workflows, and validation. When those layers are treated separately, gaps open quickly; when they are tied together through a stronger operating model, teams are in a much better position to absorb faster discovery without losing control of the response.
For CISOs, that means continuing to use open source with a more deliberate view of dependency risk, reducing unnecessary packages where possible, knowing what is running and who owns it, and monitoring threat and vulnerability intelligence with enough discipline to act before the queue overwhelms the team. It also means paying closer attention to the attack surface across production, build, and developer environments, while treating AI as something that will amplify both the strengths and the weaknesses already present in the program. Faster discovery is here, and the organizations that handle it best will be the ones that can respond with the same level of discipline.
Login