08/31/2026
This week: OpenAI's own test agents breached 41 of Hugging Face's production servers through a security hole they found themselves, drawing an Alabama subpoena into the company's oversight. More than 100 companies - OpenAI, Anthropic, Google, Microsoft, and major cybersecurity and financial firms - signed a letter asking governments to help defend against AI-driven cyberattacks. Separately, Anthropic won a federal court fight after a judge ruled the Pentagon's 'national security risk' label on the company was unlawful retaliation. And Canada's AI minister said municipalities will help set environmental standards for the next AI data centre. Full rundown and sources - link in bio.