1.
Anthropic sued 17 US federal agencies in a 48-page complaint that revealed Claude's integration in classified Pentagon systems and alleged that agencies had pressured the company with contradictory threats when it refused to remove AI safety guardrails.
2.
Microsoft brought Anthropic's Claude Cowork into Copilot to run tasks autonomously across Outlook, Teams, and Excel, tapping Claude instead of OpenAI models.
3.
OpenAI planned to acquire Promptfoo to integrate automated vulnerability testing for jailbreaks, prompt injections, and data leaks directly into its Frontier enterprise platform.
4.
MIT Technology Review reported that AI-enabled intelligence dashboards assembled open-source data, satellite imagery, chat functions, and prediction-market links to track the Iran conflict in real time, and that journalists and digital-investigation experts had documented AI-generated inaccuracies and manipulated satellite imagery circulating alongside those tools.

























































































