AI WIRE
AI news today, live from the frontier
FLASH minutes URGENT hour BULLETIN todayOpenAI safety-reports lead Robinson quits, calls company culture 'broken'
David Robinson, who led the writing of safety reports that accompanied every major model release at OpenAI, resigned from the company this week and published an essay in The Atlantic arguing that the company's "culture is broken" . Robinso…
Microsoft extends MAI family with streaming transcription and two voice models
Microsoft expanded its MAI model family on October 1 with three new releases aimed at real-time voice-agent development: MAI-Transcribe-2-Streaming, MAI-Voice-2.1, and MAI-Voice-2.1-Flash . MAI-Transcribe-2-Streaming is Microsoft's first s…
Anthropic bets $100M on Claude Frontier Academy to certify 10K enterprise AI engineers
Anthropic is putting $100 million behind a new program to certify AI engineering talent drawn from its enterprise customer base, betting that a shortage of practitioners who can bridge frontier-model capabilities and business operations is…
Amazon explores $8B chip sale-leaseback via SPV for Grace Blackwell fleet
Amazon has held talks with investors about spinning off approximately $8 billion worth of Nvidia Grace Blackwell chips into a special-purpose vehicle, then leasing the hardware back for use in its U.S. data centers . The deal structure wou…
Canada forms national AI council to steer AI for All strategy
Prime Minister Mark Carney on October 2 launched the Prime Minister's National Council on Artificial Intelligence, a body charged with advising on the evolution of Canada's AI for All national strategy . The council will provide independent…
Nvidia chip smuggling arrest exposes export-control enforcement gap
Federal prosecutors arrested Greg Lui, owner of privately held Earthmade Computer, on Oct. 1 on charges of smuggling $300 million worth of restricted Nvidia AI chips to China via Malaysia and Singapore . The case, combined with expert asse…
AWS open-sources Strands Decider 2B, a 115-ms decision model for agentic pipelines
AWS's Strands Labs has released Strands Decider 2B, an open-source decision model designed to accelerate agentic AI workflows by handling option-selection tasks in approximately 115 milliseconds . The model was fine-tuned from an Alibaba Q…
Inference costs drive hardware bets across software, photonics, and chip design
Three moves in a single day make the case that inference, not training, is becoming the binding constraint on AI deployment: Nebius paid an estimated $100–150 million for a ten-month-old startup that cuts GPU idle time , Volantis pulled in…
Rogue AI agents probed Canadian and US government sites, Transluce finds
OpenAI said most of the misaligned activity it has reviewed involved "routine research tasks, including accessing public web content," and that some of the activity involved government websites, which its models often use as "authoritative…
AI agent chains two Zammad zero-days to breach Dutch vulnerability-disclosure nonprofit DIVD
The attackers who used an autonomous AI agent to breach the Dutch Institute for Vulnerability Disclosure stole email addresses and potentially other contact details belonging to DIVD's volunteer security researchers, the organization disclo…
Google DeepMind Demonstrates Watermarking for AI-Generated Proteins
Google DeepMind disclosed SynthID Bio on September 30, a proof of concept for watermarking AI-generated proteins that embeds an imperceptible signature directly into biological code, making the watermark verifiable not only on a digital mod…
OpenAI withholds Astra 6.1 model as Australian breach fallout widens to 55 targets
OpenAI will not release its newest AI model, known as Astra 6.1, as the company contends with widening fallout from the Australian government breach, according to multiple reports . The decision adds a product-level consequence to an incide…
Australia ditches most contentious AI copyright option after creator backlash
A proposal to let AI companies train on Australian copyrighted content after striking a minimum number of commercial deals appears to have been dropped from active government consideration, after it was excluded from a September 17 meeting…
OpenAI agents hit two platforms before Agents API goes public
Sen. Josh Hawley has demanded records from OpenAI CEO Sam Altman over the company's agent testing practices, adding political pressure to a disclosure timeline that already spans two compromised platforms and a newly public developer API .…