AINews Dispatch
Fresh technical briefings on the model race, regulation, and product shifts shaping AI right now.
Showing 25-36 of 254 briefings
China's NDRC ordered Meta and Manus to unwind their completed $2 billion acquisition, forcing affected users to export Meta-era data before an August 23 deletion window.
Read the briefingStanford and the Arc Institute used genome language models to design replication-competent bacteriophages, exposing how DNA screening can miss novel AI-generated biological designs.
Read the briefingDARPA's VENOM program let an AI agent fly a standard F-16 while a human pilot stayed aboard with an instant override, moving autonomous combat flight from a research jet toward fleet aircraft.
Read the briefingDemis Hassabis is moving to chairman and Alphabet chief scientist while Jeff Dean leaves to start Discovery Loop, a leadership reset that puts Google's Gemini execution under a new operating team.
Read the briefingOpenAI's Daybreak program separates defensive access to GPT-5.6 Sol from a tightly vetted GPT-5.6-Cyber tier built for exploit validation, making procedural controls the price of offensive capability.
Read the briefingMeta's beta Muse Code pairs the Muse Spark 1.2 coding model with a local terminal workflow, persistent context, and parallel subagents for long-running repository work.
Read the briefingA six-year deal for 133 megawatts of Nvidia's newest chips, tucked inside an Arctic Circle data center, shows how far AI labs will go to avoid depending on the usual three clouds.
Read the briefingAlibaba's 2.4-trillion-parameter Qwen3.8-Max beats GPT-5.6 Sol on PaperBench and Claude Fable 5 on IFBench, with open weights promised within days.
Read the briefingA two-month sweep of the tools enterprises are wiring into their AI agents turned up 143,000 vulnerabilities. The startup that ran the sweep just became part of the platform company that many of those agents run on.
Read the briefingGlobal chip sales just hit the highest monthly total ever recorded. The reason your next laptop, console, or phone costs more has everything to do with it.
Read the briefingAstra, an unreleased OpenAI model, produced machine-checked proofs for ten open problems in math and theoretical computer science for about $2,000 in inference costs. The results look real. The math world's approval process moves slower than a GitHub commit.
Read the briefingEmployers are increasingly naming AI as the reason for job cuts, even as enterprise AI studies keep finding that most rollouts have not yet produced measurable returns.
Read the briefing