Government agencies, cybersecurity companies and threat researchers are pouring resources into studying how fast-developing AI tools can be wielded by malicious actors to hack into victim organizations.

But as agentic AI becomes more embedded in business infrastructure, there’s also a high possibility that a breach could be caused by an insider guiding the tool, whether maliciously or due to lack of security controls.

In research shared exclusively with CyberScoop, DTEX researchers detail how a common workflow in Anthropic’s Claude Cowork used in corporate environments offers convenience for AI agent deployment but grants near-total access to the system.

Claude Cowork includes tools that let users remotely control their agents. One particular tool, known as Dispatch, relays commands from a user’s phone to their desktop Claude agent. It also includes a plugin for communicating with Salesforce AI agents that access and transfer data.

DTEX researchers tested two scenarios. The first prompted Claude to summarize information from Salesforce and paste it into a draft Outlook email. The second tasked the agent with archiving selected files and transferring them via the Cowork app.

In both cases, researchers used simple, single-turn prompts and spent between 10-30 minutes preparing to exfil  the data.

Alex Desmond, director of insider threat intelligence and innovation at DTEX, told CyberScoop that both improvements in frontier models and deeper integration of AI tools into IT network operations have reduced the time defenders have to react to a breach.

“In cyberattacks, you talk about the kind of execution time of adversaries coming in and dropping ransomware, we’re now seeing the kill chain drop to 30 and 10 minutes depending on what they’re doing,” Desmond said. “Six months ago, that was a couple of hours.”

But that speed, when paired with direct access to business networks or cloud services, can also create an insider threat nightmare for organizations that must monitor for both malicious actors and potential mistakes from legitimate employees using the technology.

Over the past few years, western IT and cybersecurity businesses have been inundated with job applicants secretly working on behalf of the North Korean government. Their salaries are used to evade international sanctions and fund Pyongyang’s nuclear program, but it also positions the individuals to access or steal sensitive data or assets from these companies. 

“You’ve got a nation-state actor getting into an environment legitimately,” Desmond said. “Now if you gave them access to AI tools on top of that…you’re like ‘here’s the keys to everything and here’s this awesome tool that’s just going to make your job – stealing our data – easier.’”

Tests by DTEX confirmed that the agents indeed had access to sensitive systems, applications and data – including the ability to download SharePoint corporate data, production documentation in OneDrive, access to Outlook email, Salesforce data (and all the data it can access), and any other files on the user’s endpoint device. For each of these applications, Claude Cowork has a dedicated plugin or API to share externally if prompted.  

To be clear, DTEX’s research does not involve exploiting a software bug or configuration vulnerability, and it doesn’t come with a CVE. It’s more of an IT governance and visibility problem. Businesses are racing to integrate AI tools into their workflow and pushing employees to use the technology while failing to put in place the kind of security controls, access policies and monitoring required to spot problems.

For instance, it may not be possible to determine how a data breach or leakage involving an AI agent actually occurred if an organization is not logging and auditing its prompts – or whether the incident was the result of an agent running amok or responding to potentially malicious instructions.

While network and cloud monitoring can identify when data is being accessed or downloaded from SharePoint, that may not be a strong enough signal to stand out for defenders.

“If a user’s normal workflow is to pull sensitive files down to work locally all the time, you don’t have endpoint monitoring and you introduce an AI agent, it then just has access to all that data” along with the ability to exfiltrate it,” Desmond said.

The post Your AI agent could become your biggest insider threat  appeared first on CyberScoop.

A Google security engineer was arrested in New York and charged with crimes related to bets he allegedly placed on Polymarket using confidential information he pulled from Google systems, the Justice Department said Wednesday. 

Michele Spagnuolo, a 36-year-old Italian citizen who lives in Switzerland, is accused of placing multiple trades on the prediction marketplace last year that netted him a profit of more than $1.2 million. He allegedly abused internal access to Google’s nonpublic Year in Search data and placed a series of bets on the most searched people on Google in 2025.

“Today’s charges reinforce a decades-old message: corporate insiders cannot use confidential information to turn a profit in our markets,” Jay Clayton, U.S. attorney for the Southern District of New York, said in a statement Wednesday. “Insider trading compromises the integrity of our markets, and the American people want this greed-driven conduct investigated and prosecuted.”

Spagnuolo was charged with violating the Commodity Exchange Act, wire fraud and money laundering, which carry a combined maximum sentence up to 50 years in prison. 

He was also served with a civil complaint by the Commodity Futures Trading Commission that accused him of insider trading. The government agency is seeking restitution, disgorgement, civil monetary penalties, trading and registration bans and a permanent injunction against further regulation violations. 

Spagnuolo has been employed as a security engineer at Google since 2014, where he built products, specifications and led multiple projects in the information security unit, according to his company bio, which has since been taken down. 

A Google spokesperson said the company is working with law enforcement on its investigation. “The employee accessed our marketing material using a tool available to all employees, but using such confidential information to place bets is a serious breach of our policies,” the spokesperson said in a statement. “We’ve placed the employee on leave and will take the appropriate action.”

Spagnuolo did not respond to a request for comment.

In a complaint unsealed Wednesday, a federal investigator said Spagnulo, who used the “AlphaRaccoon” user name on Polymarket, took deliberate steps to conceal his use of nonpublic information, including efforts to obscure the source and ownership of his proceeds. 

Prosecutors noted that Google’s internal software tool, which provided Spagnuolo access to search trends, bore a banner that stated “Google Confidential” in red text, adding that Spagnuolo confirmed he understood the company’s various confidentiality and ethics policies to access the data. 

Spagnuolo allegedly created his Polymarket account in May 2024 and placed a series of trades between later that year risking approximately $2.75 million on 25 outcomes that the market treated as unlikely.

The FBI said it traced Spagnuolo’s Polymarket account to a cryptocurrency wallet he allegedly used to fund the account and initiate multiple transfers. Spagnuolo is also accused of sending multiple transactions through a cryptocurrency swapping service that were received by an account in his name linked to his Italian government ID card. 

Spagnuolo allegedly changed his Polymarket username to an alphanumeric wallet address in early December, after Google released its Year in Search results and multiple users on Discord and X speculated the person between the account was a Google insider.

The post Google security engineer accused of turning confidential search trends into $1.2M win on Polymarket appeared first on CyberScoop.

Two New Jersey men were sentenced Wednesday for facilitating North Korea’s long-running scheme to plant operatives inside U.S. businesses as employees, generating more than $5 million in illicit revenue for the regime, the Justice Department said. 

The U.S. nationals — Kejia Wang, also known as Tony Wang, and Zhenxing Wang, also known as Danny Wang — were part of a years-long conspiracy that placed operatives in jobs at more than 100 U.S. companies, including many Fortune 500 companies, based in 27 states and the District of Columbia.

The elaborate scheme involved shell companies posing as software development firms, money laundering, and espionage with national security implications. Operatives involved in the conspiracy stole sensitive files from a California-based defense contractor related to U.S. military technology controlled under International Traffic in Arms Regulations (ITAR), officials said.

“Democratic People’s Republic of Korea (DPRK) IT workers are not limited to revenue generation. When tasked, they can operationalize their placement and access to support strategic intelligence requirements, including intellectual property theft, network disruption or extortion,” Michael Barnhart, nation state investigator at DTEX, told CyberScoop.

While most of North Korea’s scheme is focused on revenue, it sometimes applies a dual-use approach, tasking certain privileged IT workers with malicious activity aiding other state-backed hacking groups, Barnhart added.

“Not all IT workers can be hackers but every North Korean hacker can or has been an IT worker,” he said. “This distinction matters for insider‑threat analysis because unlike typical fraudulent hires motivated by personal financial gain, IT workers can inflict national‑security‑level damage.”

Kejia Wang, 42, Zhenzing Wang, 39, and their co-conspirators stole the identities of at least 80 U.S. residents to facilitate the hiring of North Korean operatives and collected at least $696,000 in fees combined, officials said. U.S. victim companies also incurred legal fees, remediation costs and other damages and losses exceeding $3 million. 

Both men previously pleaded guilty to an assortment of crimes. Kejia Wang was sentenced to nine years in prison for conspiracy to commit wire and mail fraud, money laundering and identity theft. Zhenxing Wang was sentenced to 92 months in prison for conspiracy to commit wire and mail fraud and money laundering. 

The pair were also ordered to forfeit a combined $600,000, of which two-thirds has already been paid, officials said.

The conspiracy, which ran from at least 2021 through October 2024, relied in part on shell companies — Hopana Tech, Tony WKJ and Independent Lab — the men set up to create the appearance of legitimate businesses. 

“Pairing a U.S. person, a U.S. address, and a front company such as Independent Lab, the facilitators created the illusion of a legitimate domestic effort allowing the IT workers to present themselves as U.S.-based without triggering suspicion during onboarding or daily workflows,” Barnhart said. 

“Front companies can act as that middle financial flow from victim companies back to DPRK units, which then pushes funds upward through the Workers’ Party of Korea to support whichever program the unit was aligned with, whether weapons development or domestic priorities,” he added. 

These front companies reflect a higher level of tradecraft that exploits a weak spot in insider risk assessments because threats aren’t always a malicious person trying to break into a network, Barnhart said. “Sometimes it looks like an entire company appearing clean on paper.”

Authorities have responded to North Korea’s scheme by targeting U.S.-based facilitators who provide forged or stolen identities and laptop farms for North Korean operatives, and seizing cryptocurrency linked to theft. 

Law enforcement wins are stacking up, but researchers warn that North Korea’s operation is massive and consistently evolving. 

The sentencing of Kejia Wang and Zhenxing Wang comes less than a month after a trio of American men were sentenced for similar crimes, including the operation of laptop farms, wire fraud and identity theft. 

The Justice and Treasury Departments have also issued indictments and sanctioned people and entities allegedly involved in North Korea’s effort to send thousands of specialized technical professionals outside of the country to secure jobs under false pretenses and funnel their wages back to Pyongyang.

You can read the full indictments against Kejia Wang and Zhenxing Wang below.

The post US nationals sentenced for aiding North Korea’s tech worker scheme appeared first on CyberScoop.

A 27-year-old North Carolina man was found guilty of six counts of extortion for a series of crimes he committed while working as a data analyst contractor for a D.C.-based international technology company, the Justice Department said Thursday.

Cameron Nicholas Curry, also known as “Loot,” stole a trove of corporate data, including sensitive employee and compensation information, which he used to extort his employer, according to court records. Curry ultimately made off with approximately $2.5 million from the victim organization in January 2024.

The insider attack underscores immeasurable risks companies accept when employees, or contractors placed in roles by a third-party recruitment company, as was the case with Curry, are allowed to access sensitive data on a company-owned laptop. Officials did not name the company.

Curry used his access to the company’s network to remove corporate data for extortion while he worked for the company between August and December 2023. Immediately following his last day of employment with the company, Curry started sending threatening emails to its employees and demanded a ransom to not leak and destroy the data.

Officials said he sent more than 60 emails to various employees and executives over a six-week period, threatening to disclose the company’s payroll data, claiming it showed significant pay inequity across the workforce. In those emails, Curry framed the data theft extortion attack as an effort to implement salary transparency.

“Loot and our partners aim to ensure that everyone is being paid accordingly, providing employees with the leverage they deserve while also adhering to federal government regulations on protected acts,” Curry wrote in one of the emails, according to the indictment.

Curry included attachments with the emails containing screenshot images of spreadsheets listing the personally identifiable information of company employees. Officials said he also warned the company he would provide employees instructions on how to address pay discrimination through mediation, the Equal Employment Opportunity Commission or a class-action lawsuit.

Some of the extortion emails got personal, including a claim that one person on the legal team wasn’t getting a bonus while most employees in high-level positions did receive bonuses. Curry also threatened to report the breach to the Securities and Exchange Commission, citing rules that require public companies to disclose cyberattacks quickly. 

The publicly traded company notified the FBI of the breach on Dec. 14, 2023 and paid Curry’s ransom demand almost a month later.

Multiple operational security mistakes helped authorities identify and build a case against Curry rather quickly. He used personal and verifiable data to establish a new Coinbase account, and two of the debit cards linked to the account Curry established to receive a ransom belonged to his mother and sister.

Authorities searched Curry’s apartment, digital devices and vehicle in Charlotte, North Carolina, just weeks after the ransom was paid. He was arrested and released on bond in late January 2024. 

Officials said Curry initiated his extortion scheme after he learned his contract with the company wouldn’t be renewed. He faces up to 12 years in prison at sentencing.

You can read the full indictment below.

The post North Carolina tech worker found guilty of insider attack netting $2.5M ransom appeared first on CyberScoop.