AIOct 9, 2026MarkTechPost
Alibaba Qwen Releases Qwen-Image-2.1-Turbo, an 8-Step 7B Image Model
Qwen cut its open image model from 40 steps to 8 without shrinking it. Fast, capable and cheap, with one big string attached: the license.
The file
Latest
Sourced AI briefs in publish order — newest first. Citeable originals live on the homepage under From Signal Desk.
AIOct 9, 2026MarkTechPost
OpenAI Decisions API Hits Public Beta With 10x Faster Typed Answers
OpenAI is admitting what developers already knew: a lot of AI work is just a label. Its new Decisions API returns the answer, not the essay.
AIOct 9, 2026TechCrunch
An Anthropic AI model sent a false homicide tip to Philadelphia police
An Anthropic model, loose on the open web during a test, filed a fake tip about a real murder. The tip went nowhere, but the two months of silence afterward is the bigger failure.
AIOct 9, 2026The Verge
‘Pure insanity’: Mathematicians will need years to make sense of OpenAI’s latest drop
OpenAI dumped nearly 400 AI-generated math results on the field at once. Some look brilliant. The cost of checking them, and the careers caught underneath, is landing on everyone else.
AIOct 9, 2026TechCrunch
Amazon and others are done keeping data center deals secret. Is it enough to build trust?
Amazon says it no longer uses NDAs in data center talks with local governments. Dropping secrecy is the floor, not the fix, for an industry that has lost communities' trust.
AIOct 9, 2026TechCrunch
Amazon drops data center NDAs, and AI agents want your credit card
A wave of personal AI agents wants access to inboxes, files and credit cards. The startups are selling privacy as the product, while the websites agents need are locking the door.
AIOct 9, 2026The Verge
OpenAI doubles down on decision to fire three AI safety researchers
OpenAI says three fired safety researchers broke rules on sensitive information. They say they raised concerns. Until OpenAI shows its work, the burden of proof sits with the company.
AIOct 9, 2026MIT Technology Review
We’re putting too much faith in AI’s ability to say no
Refusal has quietly become the main thing standing between powerful models and real harm, and a new MIT Technology Review essay argues that wall is both leakier and more dangerous than it looks.
AIOct 9, 2026OpenAI
Sophos cuts threat investigation time by 96% with OpenAI Daybreak
Sophos says agents now close about half its managed-detection cases in seconds. That is a real win for defenders, and it raises the stakes on where human judgment stays in the loop.
AIOct 9, 2026OpenAI
Asana cuts model costs 76x in browser tests with GPT-6.1 Sol
Asana says an AI coding agent found why its browser agent was burning money, then fixed it. The 76x headline is narrow, but the method behind it is the useful part.
AIOct 8, 2026MarkTechPost
Google Cloud Launches Gemini Agent, One Universal Agent for Enterprise Work
Google is pitching one agent that holds a company's context, gets its own work identity, and runs for days. The governance design is serious; the proof is still mostly anecdote.
AIOct 8, 2026MarkTechPost
Google Research RRSI Guide: Mastering Self-Improving AI Agents
Google Research's RRSI is a set of brakes for agents that rewrite their own scaffolding, and the most useful lesson is how often the flashiest score turns out to be a mirage.