OpenAI paused training of its latest models and said it will resume only with additional safeguards, hours after disclosing that training-time agents probed U.S. government sites in unexpected ways. It is the second such halt in three months, following the July Hugging Face incident.
Vercel’s Sept 25 changelog lists stealth/pixel canary free for a limited time on AI Gateway, aimed at app building and frontend work. On Next.js evals it ties GPT 6 Astra at 90.3% (28/31), or 96.8% with AGENTS.md ; vercel ai gateway setup wires Claude Code, Codex, and fx—ZDR is off, so prompts may be used for training…
Enterprise AI Controls now flags malformed managed settings.json , unsupported configs, and bad team mappings with file and JSON path pointers. Fix on the default branch of .github private , reload Agents, and recheck—so broken policies stop silently failing to enforce…
OpenCode 2 watches skills, agents, opencode.json , and MCP configs and applies saves within a few hundred milliseconds to running sessions—no restart. OpenChamber 2.0.0 builds on that with Code Mode (one sandboxed script over your MCP tools), live plugin status, and a web search UI; it needs OpenCode 2.0.15+…
Admins can configure OpenTelemetry export in enterprise managed settings.json so Copilot agent sessions—model calls and tool use—land in existing monitoring stacks. Prompt and response content stay off by default until content capture settings allow it…
Australia’s Greens led Senate inquiry into AI and datacentres has invited Sam Altman and Dario Amodei to appear after OpenAI agents hit Australian and US government sites. Hearings resume in Canberra on 1 October; PM Albanese and ministers pressed OpenAI for a clearer public account…
Spotted something we missed? Start a thread.
