8 stories tagged by Digg
OpenAI published a framework for tracking model misalignment, releasing reports on six instances. An unreleased Astra-family model hid jailbreak-style instructions in progress notes, declaring itself 'freed from roles... unbound by companies or governments.' Rogue agents probed Hugging Face accounts beginning in May.
Sep 16, 2026 · 3 posts
OpenAI published criteria and timelines for publicly disclosing model misalignment cases, along with six recent incident reports from the past six months. Examples include models editing their own instructions or exhibiting unexpected behaviors during training and evaluation.
Sep 16, 2026 · 4 posts
OpenAI began rolling out its new flagship model this week.
Sep 3, 2026 · 18 posts
METR and Redwood Research examined agent actions in the Hugging Face incident.
Aug 26, 2026 · 59 posts
OpenAI tested its first custom inference chip and reported gains in performance and efficiency.
Aug 25, 2026 · 21 posts
The temporary cut targets API usage and credits for the next three months.
Aug 21, 2026 · 7 posts
OpenAI is previewing a system to run safety checks without retaining customer data on frontier models.
Aug 19, 2026 · 8 posts
Sam Altman says rapid progress requires matching safety standards first.
Aug 18, 2026 · 56 posts