- Describes GPT‑6 Astra as the company’s most capable and aligned model to date.
- Says an internal, more capable model proposed a solution to the Navier–Stokes problem (presented as an example of accelerated discovery).
- Frames capability gains alongside increased emphasis on safety monitoring and potential slowdowns in deployment when risks are unacceptable.
- Announces partnerships to strengthen European cyber defenses and a programme to help Ukrainian news organisations access AI tools.
- Says OpenAI awarded grants totalling $1 million and up to $1 million in API credits to selected projects responding to a call for proposals.
- Mentions ChatGPT for Teens and research funding focused on teen development and policy work.
- Argues for shared international standards to assess capabilities, manage risks and preserve human control.
- Says OpenAI will slow or stop development when proceeding would present an ‘unacceptable safety risk.’
- Calls for compatible testing, independent assessment, incident reporting rules and stronger cybersecurity protections.
Useful?