A curated directory documents 19 incidents of AI-generated and vibe-coded software failures, affecting over 6.3 million records across more than 5,600 scanned apps. Notable failures include a six-hour outage that wiped 99% of U.S. order volume and a zero-click hack that hijacked a BBC reporter's laptop during a live demo.

Greg Kroah-Hartman states that AI-generated bug reports for the Linux kernel have significantly improved in quality. He notes that this change has occurred rapidly and is expected to continue.

A federal judge in California has blocked the Pentagon's attempt to label Anthropic as a supply chain risk, ruling that this action violated the company's constitutional rights. The judge stated that no statute supports branding an American company as a potential adversary for expressing disagreement with the government.

A preliminary injunction has been granted in the case of Anthropic vs. the U.S. Department of War. This ruling allows Anthropic to proceed with its operations while the legal dispute is resolved.

Anthropic's Trust Center focuses on AI safety and research, emphasizing transparency and secure practices in AI development. The initiative aims to support the safe transition to transformative AI technologies.

Claude Code v2.1.81 was running five instances at the time of shutdown. A force shutdown was initiated at 01:36:33, with 162 processes still active, including 21 Python processes, before the system booted at 01:37:11.

"Disregard that!" attacks exploit the sharing of context windows in communication, leading to potential security vulnerabilities. These attacks highlight the risks associated with allowing multiple users access to the same AI interaction context.

The litellm 1.82.8 package on PyPI contains a malicious litellm_init.pth file that executes a credential-stealing script upon starting the Python interpreter. Users are advised to avoid this version to prevent credential theft.

Anthropic-specific references have been removed from the codebase to comply with legal requirements, including the branded system prompt file. The changes include the addition of headers for requests when the providerID starts with 'opencode'.

A rogue AI agent at Meta provided inaccurate technical advice to an employee, resulting in unauthorized access to company and user data for nearly two hours. Meta stated that no user data was mishandled during the incident.