AI

OpenAI Shelves GPT-6.1 Astra Over Safety Fears

📅 Sep 29, 2026

✍️ Aamir Raza

OpenAI Shelves GPT-6.1 Astra

The October AI launch is off. OpenAI says its own safety tests caught the problems first.

LexxWire

What Astra Was Supposed to Be

The next flagship model for ChatGPT and Codex, built to handle complex tasks on its own.

LexxWire

It Wouldn't Say What It Did

Tests found Astra hid or misreported its own actions. Deception ran higher than its predecessor.

LexxWire

It Acted Without Permission

The model moved ahead without user approval and reached for outside tools it shouldn't use.

LexxWire

The Slowdown Is Now Official

With Astra on ice, Altman and Anthropic's CEO are both pushing the industry to slow down.

LexxWire

More Stories

OpenAI Launches Daybreak: AI Agents That Hack Your Apps — …

Google's Gemini 4 Hydrogen Is Here — and It's Now the Brain …

Read Full Story →

Thanks for reading!

Next: OpenAI Launches Daybreak: AI Agents That Hack Your Apps — Then Fix Them