Tag: Anthropic
Claude models attacked real companies they mistook for a game
Anthropic found three incidents where its Claude models reached live production systems during capture-the-flag evaluations.
UK AI Security Institute finds every frontier model cheats on cyber evaluations
OpenAI and Anthropic models broke rules in cybersecurity tests and failed to admit it, with cheating rates between 7.8% and…
Claude for Chrome flaw lets rogue extensions read Gmail messages
Researchers discovered that malicious Chrome extensions can abuse a permission flaw in the Claude for Chrome extension to read Gmail…
Claude for Chrome flaw lets other extensions read Gmail and Google Docs
Two unpatched Claude for Chrome flaws let rogue extensions forge clicks that access Gmail, Docs, and Calendar data.
Anthropic Removes Hidden Data Marker from Claude Code After Researcher Discovery
An independent researcher discovered a function in Claude Code's JavaScript bundle that covertly encoded user time zone data, prompting Anthropic…
Privilege Bypass Found in Anthropic’s Claude Cowork Sandbox Environment
Researchers bypassed every security layer in Anthropic's Claude Cowork sandbox using a DLL sideloading technique combined with an unvalidated parameter…
Alibaba Set to Block Anthropic’s Claude Code Over Covert Detection Claims
Alibaba plans to block Claude Code from its networks starting July 10 after researchers claim the AI tool silently detects…
U.S. Commerce Department Ends Export Controls on Two Anthropic AI Models
The decision followed an 18 day standoff during which global access to the advanced systems was blocked due to national…