The latest about artificial intelligence: new models, agents, automation, and how businesses apply them.
Hackers breached the Pentagon and FBI within a month. Data on 2.8 million service members leaked. Here's what happened and why it matters to everyone.
OpenAI launched a series on the AGI economy. The key idea: brilliant machines matter most for routine work, not breakthroughs.
Docker proposed the Apache 2.0 Sandbox Kit v3 spec to CNCF, packaging an agent, tools and permission requests into a standard OCI image.
Introduced on Sept. 30, Gemini 4 Argon is Google's first flagship since Gemini 3 Pro and rolls out first through the Fairwind cybersecurity program.
Microsoft released MAI-Transcribe-2-Streaming at 54 cents per audio hour with 320-millisecond hypotheses, plus two text-to-speech models.
Nvidia launched its Open Agent Safety Platform with OpenShell and Sentry, joined by over 100 organizations including Microsoft and Salesforce.
TechCrunch Disrupt runs October 13-15 at Moscone West with 200 sessions, six stages and Startup Battlefield for 200 teams competing for $100,000.
OpenAI confirmed dismissal of three safety researchers after an internal probe found mishandling of sensitive research outside established procedures.
Cerebras reported a 5x inference throughput gain with the same hardware footprint by splitting prefill and decode into separate pools.
Graphite compared AI rewrites with 10,000 human articles and found 13,000 tells, from dependable to this matters.
Amazon Payments ran a seven-week test of multi-objective contextual bandits on SageMaker AI, reporting a high single-digit lift in final conversion for one audience.
Arceus Legal raised $17M led by Greycroft to expand its flat-fee legal service, where most contract reviews finish in three to five hours.