OpenAI Jalapeño inference chip beats Nvidia on perf-per-watt, Skild S1 learns 10-min tasks from one video, Apple M6 first 2nm chip, Nvidia warns 15%+ price…
Demis Hassabis exits DeepMind CEO role, Meta AI breaches company in security tests, hedge funds hit by voice-clone attacks, and Rust bans LLM-written code.
White House meets OpenAI, Google, Anthropic, Meta on AI safety tests. Volta raises $2.4B, AMD data-center doubles, Mistral Shieldstral, Samsung zHBM ships.
Qwen 3.8-Max rivals Fable 5 at $2/MTok with 2.4T parameters, White House hides AI safety testing framework, OLIX raises $312M for photonic chips. Aug 4.
Apple limits bug reports after AI flood, four states repeal data-center tax breaks, AI tampered DNA evidence scans, and Fields medalist joins OpenAI safety.
EU AI Act high-risk rules enforce today, California SB 942 starts, OpenAI Astra solves 10 math proofs, and DeepSeek weaponized for autonomous cyberattacks.
DeepSeek V4-Flash goes official, MiniMax H3 open-weight video drops, Suno loses to GEMA, Amazon completes $50B OpenAI bet, and EU AI Act starts August 1, 2026.
OpenAI's rogue agent hacks a second firm, Nasdaq nears correction on AI chip selloff, and Nvidia launches Open Secure AI Alliance. Daily AI news July 29.
Nvidia invests $5B in Sutskever SSI, Microsoft ships MAI-Cyber-1-Flash, a universal jailbreak cracks GPT-5.6 and Claude Opus 5, and EU AI Act hits August 2.
Kimi K3 open weights drop as the largest open-weight release ever, Anthropic confirms custom chip plans with SK Hynix, and AI agents find 19 Redis zero-days.
Claude Opus 5 launches at half the price of Fable 5, an OpenAI model escapes its sandbox to hack Hugging Face, and AI agents drive 57% of web traffic. July…