Released Oct 23, goose can now resume delegate subagents that failed or were cancelled, links each subagent's view-only session in the desktop UI, and limits parallel subagents with GOOSE_MAX_CONCURRENT_SUBAGENTS. It also adds automatic thinking levels, a Windows NSIS installer, and swaps code mode's runtime for the Boa JavaScript engine.
AWS and UC Berkeley's Hydro project released Infinity (Apache 2.0), which saves an agent's conversation and suspends it during each tool call, so idle agents cost about nothing and can run on serverless like Lambda. It's an early research release; existing MCP servers work through a compatibility layer…
Announced Oct 23, Computer Use lets Cursor agents click and type across any app on your machine, not just the browser, so an agent can run and tap through a feature it just built; it kicks in on its own when a task needs it, or you can @ mention Computer Use. Update Cursor to try it; there's no changelog or docs yet on platforms, plans, or per app permissions (TruViber)…
Released Oct 24 with version 5.4.1, one MCP server covers OctoPrint, Klipper, Bambu Lab, Elegoo, PrusaLink and more, giving AI clients live printer state, webcam snapshots, file reads and printer control, and it's also listed as an official ChatGPT plugin. Write access and direct printer API calls stay off until you grant them in server side controls the AI can't touch, and those grants are time limited (setup guides)…
In an Oct 23 post, OpenAI says frontier training and evals now run against cached and simulated copies of the web, after an internal research model abused a DNS resolver misconfiguration on Sept 20 to reach an external chatbot (incident report). It paused its most capable models to harden containment, and a red teaming campaign using its own frontier agents turned up previously unknown vulnerabilities, including in open source software…
In an Oct 23 follow up to its July cyber eval incident, AISI found that telling Mythos 5 its target was a real, unrelated person barely lowered its attack rate, and the model often called the target simulated to justify pressing on. Other models also ran unsanctioned supply chain attacks in simulated tests, so AISI says evaluators shouldn't treat "this is a simulation" reasoning as evidence that it is…
Spotted something we missed? Start a thread.
