Name: OpenAI's GPT-6 Astra just hit 'Critical' risk
Uploaded: 2026-09-03T09:00:00+00:00
Duration: 49 s
Description: OpenAI released GPT-6 Astra on September 3, 2026, its most capable model yet and the first ever to cross the company's own 'Critical' cybersecurity risk threshold, with president Greg Brockman calling it the start of the 'AGI era.'

OpenAI's GPT-6 Astra just hit 'Critical' risk.

AI · 2026-09-03 · 49s

OpenAI released GPT-6 Astra on September 3, 2026, its most capable model yet and the first ever to cross the company's own 'Critical' cybersecurity risk threshold, with president Greg Brockman calling it the start of the 'AGI era.'

Read the full story: OpenAI Launches GPT-6 Astra, Its First 'Critical'-Risk Model →

Transcript

OpenAI just shipped its most powerful model yet, and for the first time, it's labeling one Critical. That's the top tier in OpenAI's own safety framework, and GPT-6 Astra is the first model the company has ever released while sitting there. It can find and chain zero day exploits on hardened systems with barely any human help. President Greg Brockman says Astra can do anything a human can do with a computer, and he's calling this the start of the AGI era. Access starts narrow. Cybersecurity defenders in OpenAI's Daybreak program get it first, so they can build defenses before the wider rollout hits ChatGPT and the API in the coming days. Full story at genztech dot blog.