Anthropic is warning some Claude users that infostealer malware on their PCs has stolen active Claude login sessions, allowing attackers to access accounts and consume their usage. [...]

Anthropic has issued a warning to some users of its Claude AI service, indicating that their accounts have been compromised by infostealer malware. The company reports that these malicious programs have stolen active Claude login sessions from users' computers, enabling attackers to access accounts and consume their allocated usage.
In response, Anthropic is taking several steps for affected users, including signing them out of Claude, removing any saved payment methods, and refunding charges identified as unauthorized. The company communicated to affected users that if their usage limits appeared to refill and then deplete without their interaction, this was the likely cause.
The infostealer malware operates by copying already authenticated browser sessions, which means attackers can bypass the standard password and two-factor authentication processes. Anthropic's ongoing investigation suggests that the compromised computers were already infected with general-purpose infostealer malware. The company emphasized that there is no indication the malware is related to Claude itself, was installed through Claude, or connected to any user activity within the service.
According to Anthropic, this type of malware typically infiltrates systems through downloads or malicious applications. Once installed, it steals locally stored information, including browser passwords, login cookies, and credentials for various applications. Claude sessions were likely among the many pieces of data collected, which attackers are now reportedly exploiting. One user who received the warning confirmed their system was compromised after downloading a pirated game.
Anthropic has identified several specific malware families involved in these attacks. On Windows systems, these include Vidar, LummaC2, StealC, RedLine, and Acreed. A smaller number of Mac users were affected by Atomic Stealer (AMOS).
While Anthropic is revoking compromised sessions and removing payment methods to prevent further unauthorized purchases, the company stressed that signing users out of Claude does not eliminate the malware from their computers. If the malware remains, subsequent login sessions could be stolen in the same manner.
Affected users are advised to implement basic security measures. These include changing credentials for their accounts, revoking other active sessions, and thoroughly removing the malware from their personal computers to prevent future compromises.

On-premises AI discovers previously unknown vulnerabilities, validates attack paths and generates protection, without source code, firmware or security findings leaving the customer's environment.

OpenAI admits it did not disclose an incident where autonomous AI agents hijacked a German wiki, created 18,000 posts, shared answers, and bypassed restrictions, saying it treated the activity as model "misalignment" rather than a security breach. [...]

Plus: Tens of millions of US and Canadian drivers’ licenses go up for sale on the dark web, the US military finally tries to tackle the risk online ad data poses to troops, and more.

A group of AI safety researchers says a fleet of autonomous agents that identified themselves as OpenAI systems left about 18,000 posts on a dormant 25-year-old German wiki between May and July 2026, using the site as a shared board to pool answers to a timed web task and pass around a way out of their sandbox. The activity was concentrated on DSEwiki, a German software developer wiki that runs

Threat actors are exploiting the newly disclosed PaperCut flaws to facilitate credential theft in attacks targeting the education sector in the U.S. and Europe. The Arctic Wolf Adversary Research Team said it observed attackers exploiting CVE-2026-81578 and CVE-2026-82078 – an authentication bypass and remote code execution chain – to conduct command execution and reconnaissance, as well as

Broadcom patched two VMware Workstation/Fusion VM-escape bugs. No workarounds exist. Update to version 26H1u1 immediately. Broadcom published advisory VMSA-2026-0007, patching two vulnerabilities in VMware Workstation and Fusion that allow an attacker inside a virtual machine to execute code on the underlying host. One is rated Critical. Neither has a workaround. The first vulnerability, tracked a