OpenAI has released its Astra model to business users through the Daybreak programme, and president Greg Brockman says it qualifies as artificial general intelligence. OpenAI defines AGI as systems generally smarter than humans, while Brockman said users should decide whether Astra meets that standard. The model is trained to operate computers like a person and is described as capable of financial modelling, tax preparation and video-game development with minimal human oversight.
The launch highlights safety and regulatory stakes. Astra reached OpenAI’s critical cybersecurity threshold, meaning it can find and build working zero-day exploits against hardened systems without human help, but both versions block the most advanced cyber capabilities. A version with additional safeguards is planned for ChatGPT subscribers. US and European authorities reviewed or can evaluate such systems, but neither jurisdiction pre-cleared this release; OpenAI’s own classification remains decisive.
https://thenextweb.com/news/openai-gpt-6-astra-brockman-agi-claim-us-voluntary-prerelease-review-eu-article-92-evaluations
oxhak / Ox HaK
npub1sx…dz7z7
2026-09-03 23:06:01 UTC
Author Public Key
npub1sxexewvzysc3affq4yzzh7w8e3udyujap2vlj7t6lkdg5dvhp24q4dz7z7Published at
2026-09-03 23:06:01 UTCEvent JSON
{
"id": "f22976a556c2be1145340895a1e8c6a9310b208296866d24f185d4f1de56c630",
"pubkey": "81b26cb98224311ea520a9042bf9c7cc78d2725d0a99f9797afd9a8a35970aaa",
"created_at": 1788476761,
"kind": 1,
"tags": [],
"content": "OpenAI has released its Astra model to business users through the Daybreak programme, and president Greg Brockman says it qualifies as artificial general intelligence. OpenAI defines AGI as systems generally smarter than humans, while Brockman said users should decide whether Astra meets that standard. The model is trained to operate computers like a person and is described as capable of financial modelling, tax preparation and video-game development with minimal human oversight.\n\nThe launch highlights safety and regulatory stakes. Astra reached OpenAI’s critical cybersecurity threshold, meaning it can find and build working zero-day exploits against hardened systems without human help, but both versions block the most advanced cyber capabilities. A version with additional safeguards is planned for ChatGPT subscribers. US and European authorities reviewed or can evaluate such systems, but neither jurisdiction pre-cleared this release; OpenAI’s own classification remains decisive.\n\nhttps://thenextweb.com/news/openai-gpt-6-astra-brockman-agi-claim-us-voluntary-prerelease-review-eu-article-92-evaluations",
"sig": "114086b9c329e1c536d6e2afeeaf086fa3982e367c85567331028880c5beb9017c0624107d0e2b08a64ba854ca8d9232ac19e37ef391df3d8ea48e6ee75a71c5"
}