Amazon delivered the quarter Wall Street had been waiting more than four years for: its cloud engine roared back to life.
Rhoda AI exited stealth with a $450M Series A led by Premji Invest, unveiling its FutureVision robot foundation model.
AMD and Cerebras stitch two rival chip architectures into one disaggregated inference pipeline, claiming 5x better tokens-per-watt.
Baselayer raised $20M in Series A funding led by Koro Capital and M13 to automate fraud and risk checks for banks.
Unive rolled out ChatGPT Enterprise workforce-wide, and staff built ~1,500 custom GPTs that cut claims prep from hours to minutes.
MAI-Realtime, Microsoft's first native full-duplex voice model, surfaced unannounced in a hidden MAI Playground preview.
Alibaba's 2.4-trillion-parameter Qwen3.8-Max lands API-first, open weights promised next week, in China's frontier model race.
OpenAI has quietly upgraded the model most ChatGPT users never choose but many eventually meet: the fallback tier.
Moonshot AI confirms kimi-k2.5 and its moonshot-v1 API series retire Aug 31, 2026, pushing developers toward Kimi K3.
Frontier AI now outscores PhD experts on hard biology and virology benchmarks, reviving dual-use biosecurity fears.
The AI models that let scientists dream up enzymes never seen in nature are also among the least understood tools in modern biology.
A new Minerva benchmark evaluates 9 LLMs on 224 enterprise tasks by checking whether system state actually changed correctly.
A new class of 'self-driving labs' couples ML design with robotic build-test hardware to autonomously engineer proteins and enzymes.
A bipartisan quartet of senators wants to hand parents the keys to their children's AI chatbot accounts via the CHATBOT Act.
California legislators returned August 3 to decide which of roughly 30 pending AI bills survive back-to-back appropriations hearings.
Trump's Executive Order 14365 set the federal government on a collision course with the states over who governs AI.