GPT-6 Astra Is the First Model OpenAI Classifies as Critical for Cybersecurity
OpenAI's Preparedness Framework rates Astra Critical for cybersecurity after expert testers found real browser and OS-kernel vulnerabilities.
AI Storyline
4 items · 2 sources · 4 days
Operational story trace
GPT-6 Astra
Follow in this browser to see new updates on your Live feed.
Current stateLive · flagged Critical for cybersecurity
Latest change
Basis became the fourth customer case study since the Critical cybersecurity rating, reporting Astra finished a 50-tab tax workbook 2x faster than GPT-5.6 Sol; still no visible new access limits.
Earlier contextThe story so far
OpenAI launched GPT-6 Astra on Sep 11 and early customers delegated production work to it with fewer human check-ins. On Sep 17 OpenAI rated Astra Critical for cybersecurity after testers found real 0-days in a browser and an OS kernel.
editor-curated · source-linked
State over time
● classified Critical for cybersecurity · Sep 17adoption continues post-classification · Sep 21-28 ●
- classified Critical for cybersecurity · Sep 17
- adoption continues post-classification · Sep 21-28
THE TURN · Sep 17OpenAI classifies Astra Critical for cybersecurity after it finds real 0-daysExpert-led Preparedness Framework testing found previously unknown vulnerabilities in a browser and an OS kernel, making Astra the first model OpenAI has rated Critical on that axis.1 source · watcher updated status · show source ▾
NOW · Sep 21–28Customer adoption keeps accelerating despite the Critical rating3 sources · show sources ▾
Higgsfield AI ships new video features in a day with GPT-6 Astra
Higgsfield AI's case study: new video ad-creation features shipped in a day, four days after the Critical classification.
Harvey turns legal context into stronger drafts with GPT-6 Astra
Harvey's case study: Astra turns legal case context into stronger, more structured drafts, freeing lawyers for strategy work.
Basis completes a tax workbook 2x faster with GPT-6 Astra
Basis's case study: a 50-tab tax workbook finished 2x faster than with GPT-5.6 Sol, a vendor-reported comparison.
Day 1 Thursday, Sep 17, 2026
GPT-6 Astra Is the First Model OpenAI Classifies as Critical for Cybersecurity
OpenAI's Preparedness Framework rates Astra Critical for cybersecurity after expert testers found real browser and OS-kernel vulnerabilities.
Day 2 Monday, Sep 21, 2026
Higgsfield AI ships new video features in a day with GPT-6 Astra
Higgsfield AI's case study: new video ad-creation features shipped in a day, four days after the Critical classification.
Day 3 Wednesday, Sep 23, 2026
Harvey turns legal context into stronger drafts with GPT-6 Astra
Harvey's case study: Astra turns legal case context into stronger, more structured drafts, freeing lawyers for strategy work.
Day 4 Monday, Sep 28, 2026
Basis completes a tax workbook 2x faster with GPT-6 Astra
Basis's case study: a 50-tab tax workbook finished 2x faster than with GPT-5.6 Sol, a vendor-reported comparison.
What to watch — open questions
- Do Astra's launch-week coding and computer-use gains hold up in independent, third-party benchmarks rather than vendor case studies?
- Does Perplexity's reduced check-in frequency reflect genuine reliability, or looser production guardrails that haven't yet been tested by a failure?
- Nearly two weeks past the Critical classification and four case studies in, has OpenAI added any new access restrictions, usage logging, or eval requirements for API customers, and were the browser/kernel vulnerabilities disclosed and patched?
How this thread was built
editor wrote the arc · 2 beatswatcher 1 status change
Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.