- Treasury threatens sanctions after White House claims Moonshot distilled Anthropic’s FableFactual summary
The White House claims Chinese firm Moonshot 'distilled' Anthropic's Fable model, leading the Treasury to threaten sanctions, intensifying US debate on Chinese open models.
- Safety guardrails blocked Hugging Face's defenders, not the attacker, when an AI agent breached its systemsFactual summary
Hugging Face's incident response team found commercial AI safety guardrails blocked forensic queries during a breach by an autonomous AI agent.
- Anthropic to support FCA’s Supercharged SandboxFactual summary
Anthropic will provide Claude access to firms in the FCA's Supercharged Sandbox, assisting with AI use case development and testing.
- AI Teammates: how monday.com runs production AI agents on Amazon BedrockFactual summary
monday.com uses AI agents on Amazon Bedrock for coding assistance, reporting 50%+ increase in developer PR throughput.
- Kick-off of new GenAI pilots for public administration & Apply AI stakeholder meetingFactual summary
The European Commission is launching three new generative AI pilot projects for public administration, with an online kick-off meeting on 14 Sept 2026.
- The Correctness Illusion in LLM-Generated GPU KernelsFactual summary
Research finds common benchmarks for LLM-generated GPU kernels (KernelBench, TritonBench, GEAK) overstate correctness by using fixed-shape, small-sample checks.