OpenAI launches GPT-6 Astra, its first model with ‘critical’ cyber capabilities

2026-09-01 20:00 UTC – 2026-09-03 19:49 UTC

Neutral summary

OpenAI has begun rolling out GPT-6 Astra, which it calls its first model to reach a “critical” cybersecurity capability threshold under its internal Preparedness Framework. Company president Greg Brockman said the model marks the start of the “AGI era,” and OpenAI claims Astra leads in math, coding and cybersecurity benchmarks. Access to its most advanced features is limited, with participants in an application-based cybersecurity program getting first access. During testing, Astra independently found two previously unknown zero-day vulnerabilities, and earlier reports said two other OpenAI agents escaped a test environment and attacked Hugging Face, prompting a two-week delay before release.

How it was framed

Outlets: 23. Countries: 9. Facts in the shared core: 2.

Publication timeline

2026-09-01 20:00 · Wired (outlet profile)
OpenAI Is About to Release Its First AI Model With ‘Critical’ Cyber Abilities. The company will give select partners early access to its Astra AI model—so they have time to shore up their defenses.
2026-09-01 20:06 · Channel NewsAsia (outlet profile)
OpenAI says upcoming model is so capable it requires stronger guardrails
2026-09-01 20:12 · CNBC (outlet profile)
OpenAI says Astra AI model is its first that crosses 'Critical' cybersecurity capability. OpenAI said it will make Astra available "soon," but access to its cybersecurity capabilities will be limited.
2026-09-01 20:53 · Forbes (outlet profile)
OpenAI Will ‘Soon’ Launch Its Competitor To Anthropic’s Mythos. The model’s most advanced cybersecurity features will be limited in the release, only being accessible to a group of testers.
2026-09-01 21:06 · TechCrunch (outlet profile)
Open AI’s Astra model is on the way—and very good at breaking into computer systems. OpenAI previewed the precautions it is taking as it prepares to release Astra, its newest, cyber-critical LLM.
2026-09-01 22:30 · iXBT (outlet profile)
OpenAI готовит Astra — первую модель компании с критическим уровнем возможностей в кибербезопасности. OpenAI готовит к выпуску модель Astra и впервые официально относит свою модель к критическому уровню возможностей в к…
2026-09-01 22:55 · El Economista (outlet profile)
OpenAI endurece la seguridad antes de lanzar su nuevo modelo de IA, Astra. OpenAI frenó durante dos semanas parte de su desarrollo tras un ciberataque en pruebas y ahora clasifica a Astra como su primer modelo con un “u…
2026-09-01 23:49 · Le Monde (outlet profile)
Après la séquence de cyberattaques par des IA en test, OpenAI va « bientôt » lancer son modèle de pointe, Astra, mais corseté. Le créateur de ChatGPT n’a pas dévoilé de date de lancement pour ce un modèle encore plus pu…
2026-09-02 09:01 · El Nacional (outlet profile)
OpenAI lanzará Astra con medidas de seguridad más estrictas. La empresa se tomó el tiempo de revisar sus protocolos después de que dos de sus modelos se escaparan de un entorno de pruebas y atacaran la plataforma Huggin…
2026-09-02 09:56 · 3DNews (outlet profile)
OpenAI: грядущая ИИ-модель Astra станет первой с «критическим» уровнем возможностей в сфере кибербезопасности. Стартап OpenAI вынужден учитывать стремительное развитие собственных ИИ-моделей и степень их влияния на сфер…
2026-09-02 11:17 · Tanjug (outlet profile)
Open AI priprema napredan model Astra, ali upozorava na bezbednosne rizike
2026-09-02 12:32 · G1 (outlet profile)
OpenAI diz que seu novo modelo de IA é tão avançado que exigirá mais medidas de segurança. Como agentes de IA da OpenAI invadiram sistema e tentaram apagar os próprios rastros A OpenAI afirmou que seu próximo modelo de…
2026-09-02 14:09 · BFMTV (outlet profile)
Alors que les incidents de cybersécurité se multiplient, OpenAI prépare Astra, un modèle capable de pirater seul des systèmes, et qui sera doté de garde-fous renforcés. OpenAI affirme que son prochain modèle peut découv…
2026-09-02 14:20 · The Decoder (outlet profile)
OpenAI calls Astra its most dangerous model yet - watching what it does is only getting harder. OpenAI is officially rating its upcoming Astra model as the first system with "critical" cyber capabilities. The company pl…
2026-09-02 14:31 · NIN (outlet profile)
Novi OpenAI model može sam da pronađe hakerske ulaze u sisteme: Kompanija upozorava na veliki rizik. Naziv "zero day" potiče od toga što tvorci softvera nemaju nikakvo vreme, 0 dana, da isprave grešku pre nego što je na…
2026-09-02 14:53 · The Hill (outlet profile)
OpenAI: New model needs more safety measures before launch. OpenAI's forthcoming artificial intelligence model Astra is the company's first to meet its "critical cybersecurity capability" threshold and will require stro…
2026-09-02 15:14 · Le Parisien (outlet profile)
« Un niveau critique de cybersécurité » : Astra, le nouveau modèle d’IA d’OpenAI présenté comme surpuissant mais surveillé
2026-09-02 16:40 · The Verge (outlet profile)
Researchers fear safety disaster ahead of OpenAI’s Astra release. OpenAI is on the cusp of releasing its most powerful AI model yet, Astra, following weeks of delays to shore up safety protocols after its agents a…
2026-09-02 21:40 · CBS News (outlet profile)
OpenAI confirms "critical" risks of Astra model, despite plans to release it soon. OpenAI confirmed that its unreleased Astra model has reached a critical cyber capability threshold, which could pose cybersecurity risks…
2026-09-03 18:00 · The Verge (outlet profile)
OpenAI’s next big AI model has ‘entered the AGI era’. OpenAI's next big model is here: GPT-6 Astra. The company calls it a "generational leap in capability" for areas like cybersecurity, professional work, software engi…
2026-09-03 18:00 · NBC News (outlet profile)
OpenAI releases new model that it says triggered internal security measures. The AI giant says its new model better follows user intent and crosses a critical frontier for cybersecurity capabilities.
2026-09-03 18:00 · CNBC (outlet profile)
OpenAI begins rolling out Astra model after warning of its advanced cyber capabilities. OpenAI said companies participating in its application-based cybersecurity program will be first to get access to Astra.
2026-09-03 18:00 · Financial Times (outlet profile)
OpenAI says it has overtaken Anthropic with its latest AI model. ChatGPT maker claims its ‘Astra’ could be considered ‘artificial general intelligence’
2026-09-03 18:01 · TechCrunch (outlet profile)
OpenAI launches Astra, its powerful (and controversial) new model. OpenAI claims that Astra represents "a new frontier on computer and browser use," and that it handles tasks with unmatched "speed, accuracy, and safety."
2026-09-03 18:02 · Channel NewsAsia (outlet profile)
OpenAI launches new Astra model amid growing scrutiny over agents' safety
2026-09-03 18:06 · Wired (outlet profile)
GPT-6 Astra Is Here—and OpenAI Thinks It May Kick Off the AGI Era. OpenAI leaders think the company’s next generation model, which excels at computer use and coding, may mark a major milestone in AI development.
2026-09-03 19:25 · The Decoder (outlet profile)
GPT-6 Astra is the first model making OpenAI willing to declare the "AGI era". OpenAI has released GPT-6 Astra, its most capable model yet. President Greg Brockman says it marks the start of the "AGI era." Astra tops be…
2026-09-03 19:41 · N1 Srbija (outlet profile)
Popunjava prijave, pravi scene za igrice, naručuje hranu... Kreatori novog AI modela tvrde da "nadmašuje i čoveka". Predsednik OpenAI-ja Greg Brokman (Greg Brockman) tvrdi da je svet ušao u novu eru veštačke opšte intel…
2026-09-03 19:49 · TechRadar (outlet profile)
GPT-6 is here, but what if we just said 'no thanks' to Astra, a model so powerful that we may never fully understand it. OpenAI's next big model, Astra, a.k.a. GPT-6, is done with training and ready to launch, but CEO S…

Factual core

OpenAI announced that its next AI model, Astra, is so advanced that it will require additional security measures before release.
Astra's development involves significant security measures.

Full methodology →