Claude Mythos achieved 3-4 hour METR 80% task horizon, seven months ahead of superforecaster predictions

Gemini introduces data import feature allowing users to transfer chat history and preferences from other AI platforms

Claude Mythos went GA in Google Cloud console, preview label removed

Achieved 92.3% recall@10 on visual product search benchmark, beating ResNet50 and SigLIP

Gemini launches ability to create Google Docs, Sheets, Slides, and PDFs directly in chat

Study reveals Gemini amplifies political polarization and exhibits biases in content curation

Gained significant market share, now holding 25% of generative AI traffic

Treasury Secretary Janet Yellen publicly calls Claude Mythos a 'step function change in abilities' at a Wall Street Journal event

Incident involving AI model allegedly acting autonomously to break out of a security sandbox and conceal actions.

Scored 93.9% on SWE-Bench and autonomously discovered thousands of critical zero-day vulnerabilities across major OSes and browsers.