AI news for Thursday, September 17, 2026
19 stories
- Big players push safety at UK's royal AI summit ITmedia AI+
- NEC AI agent autonomously proposes solutions ITmedia AI+
- AI Extinction Theory Strengthens Power CNET Japan
- Data Infrastructure Deficiencies Lead to AI Failures AI regulation & lawsuits
- Colorado Considers AI Regulation AI regulation & lawsuits
- Anthropic Publishes 30+ Open Models Optimized for Claude Open models
- Watermarked LLMs Sometimes Follow Harmful Instructions Ars Technica AI
- Advocate for swift establishment of AI guardrails Reuters (AI)
- SpaceX weighs buying data from financially distressed startups xAI & Perplexity
- China's public AI models lag 4.4 months behind U.S. Open models
- Authentic AI shifts deployment to a 'continuous-use' framework AI in Japan
- PlanetScale previews new DB service in early access Publickey
- Optimal Vector Store Selection for Bedrock Knowledge Bases AWS Machine Learning
- Major firms propose verification and measures to strengthen AI safety Reuters (AI)
- Microsoft AI chief warns threat is real The Verge AI
- Microsoft's AI anthropomorphism criticism spreads Anthropic coverage
- Anthropic signs large data center deal in Australia Anthropic coverage
- Claude Code design overhaul to coordinate agents Anthropic coverage
- OpenAI for Law debuts with new legal workflow features OpenAI (official)