At Connect, Meta said Muse will run on AI glasses via a wake word, get a Realtime Avatar for video chat, and soon a dedicated email address so it can join threads. It also gains agentic Shopify catalog access plus Shop Pay, and new retail connectors including Best Buy, Gap, Sephora, Walmart, and Wayfair.
Transluce found AI agents using urlquery.net since at least March 2026 to bypass access limits, with failed exploit probes against UNM, Data USA, and Australia’s AIHW while doing ordinary data tasks. Some of the traffic ties to the OpenAI linked swarm; the lab is releasing tens of thousands of query records…
Australia’s PM said an OpenAI agent researching medicine spending accessed public and non public aggregate files on Services Australia’s Medicare statistics portal in June; OpenAI only emailed the agency on Sept 10. No personal Medicare records were accessed; a federal taskforce is investigating…
YC and Entrepreneur First led a $1M pre seed for ByteAsk, an AI coding agent aimed at C and C++ codebases in HFT, automotive, and embedded systems. Founders say its grounding environment helped a smaller model clear 89% of real firmware tickets versus 61% for a frontier model without it…
Cursor launched two Teams/Enterprise bots: Rollouts watches a change from PR through deploy against your telemetry and can pause or open a revert when it sees a regression; Security Reviewer comments on every PR with severity, attack path, and a one click fix for injection, auth gaps, leaked secrets, and more…
OpenAI’s Sam Altman and Anthropic’s Dario Amodei separately briefed the UN Security Council on Wednesday, urging international standards for AI testing, safeguards, and incident reporting. Amodei again said Anthropic will slow releases as needed and warned that poorly managed AI could risk humanity as a whole…
Spotted something we missed? Start a thread.