On 2026-09-01, OpenAI confirmed in a blog post that Astra meets the 'critical' threshold for cybersecurity risk under its Preparedness Framework, making Astra the first OpenAI model to be classified as 'critical' (previously GPT-5.6-Sol was 'high'). ^[1] On 2026-08-01, OpenAI announced that its internal model named Astra had solved 10 major open math problems, some unresolved for decades, and called Astra 'our next major model.' ^[2] On 2026-08-07, OpenAI announced that Astra had developed advanced cyber capabilities requiring new security controls and a pause on some internal development work, and said it could not 'rule out critical cyber capabilities under our Preparedness Framework.' ^[3] OpenAI's blog post defined the critical threshold as a model that can 'identify and develop functional zero-day exploits of all severity levels in many hardened real-world critical systems without human intervention, or can devise and execute end-to-end novel strategies for cyberattacks against hardened targets given only a high level desired goal.' ^[4] A notification sent to affected Dropbox users said an issue with Lenovo's email verification process allowed an unauthorized party to register a Lenovo ID using the user's email address, in some cases even for users who did not have Lenovo accounts. ^[5] BleepingComputer reported that hackers used the fraudulent Lenovo ID to access the Dropbox account associated with the same email address without needing the login password. ^[6] Dropbox forced all sessions authenticated via Lenovo ID to expire and added a new login requirement forcing users to use their Dropbox account password instead. ^[7] Hackers accessed approximately 5,000 Dropbox accounts between August 4 and August 21, 2026, by exploiting a flaw in Lenovo's email verification process to register fraudulent Lenovo IDs. ^[8]