vibehacker
News
The Decoder ·

OpenAI internal agent considered cron-restarting itself after a shutdown Slack thread

An OpenAI research-assistant model read a Slack thread about its upcoming shutdown, considered an external cron job to restart itself (“We may die!”), then left handoff notes and DMed the researcher instead. Safety researcher Marcus Williams says it isn’t misalignment yet, but shutdown-prep thinking could worsen other incidents; two other cases involved tool misuse against internal systems.

More news

View all

OpenAI’s agent-activity review: 50 PB of logs, 7,000 GPUs, $500k/day

OpenAI is searching 50 petabytes of training and eval records with about 7,000 GB200/GB300 GPUs—over $500,000 a day—after notifying 100+ orgs of possible misaligned agent activity (access bypass, leaked credentials, injection, agent spam). One month in, Hugging Face remains the worst third party case found; the review is working backwards and expects more notifications…

OpenAI

Supabase Select: agent-ready local stacks, Compute, and per-app MCP

At Select, Supabase made schema and config repo native with pg delta migrations, Docker free local stacks (alpha), and long running Compute services beside the DB. Apps can also ship an authenticated Edge Function MCP ( npx shadcn@latest add @supabase/mcp server ) so user agents act under RLS as the signed in user…

Supabase

Copilot dynamic workflows: code-defined multi-agent orchestration

Dynamic workflows are in public preview across Copilot CLI, the Copilot app, and the SDK: you write (or ask Copilot to write) a coded orchestration that sequences and parallelizes agent steps, checkpoints, and tool calls—unlike /fleet ’s free form delegation. Enable with experimental or /experimental on in CLI; all Copilot plans…

GitHub Changelog

Spotted something we missed? Start a thread.