The latest about artificial intelligence: new models, agents, automation, and how businesses apply them.
Moonshot AI introduced Kimi K3.1 Pro, a 3-trillion-parameter model that beats GPT-6 Astra on coding and finance tests ahead of open weights.
OpenAI will train frontier models on cached and simulated web content after a September 20 incident forced a pause in training work.
Harvard scholar Martha Minow told MIT that AI erodes thinking and information, and urged engineers to build tutor-like tools.
TechCrunch Founder Summit 2026 takes place November 4 in Boston with Toast, Bessemer and 7AI speakers on fit, scaling and pitches.
GitHub logged 7.38 billion commits in September 2026, and Checkmarx says security teams need deterministic checks plus AI.
OpenAI, PerturbAI and Scripps Research used close to 2,000 agents on GPT-Rosalind to analyse a whole-brain CRISPR atlas in 32 minutes.
Maxinsights raised $30 million to build a data layer for physical AI from 2 million hours of human activity footage.
Anthropic launched OSS Scanner, a free opt-in service that sends unreviewed AI security reports with reproducers and patches to critical open-source maintainers.
OpenAI signed an AI safety research agreement with Poland's NASK institute on October 23, 2026, building on cyber work that found six MikroTik flaws.
inDrive signed an MOU with Pony. ai and Nvidia to test Gen-7 robotaxis in Astana in the first half of 2027, starting with a safety operator on board.
Kazakhtelecom has deployed 880 NVIDIA GPUs in Astana and Almaty, with over 2,000 more planned for next year alongside an inDrive and Pony. ai robotaxi assessment.
OpenAI is negotiating a $30 billion bridge round at a $1.25 trillion valuation, with UAE investors leading and Coatue participating.