한국어EN
Safety, Alignment & InterpretabilityOpenAI

Preparedness Framework v2

Made by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made

Preparedness Framework v2

Why it matters

Included AI self-improvement among the Tracked Categories

Source — Primary sources first; secondary reports are labelled “Press”

Officialopenai.comhttps://openai.com/index/updating-our-preparedness-framework/

Added in Week 35, 2026 · See that week

On the map — ⑩ Safety, Alignment line

  1. Previous stopAnthropic classifiers
  2. This recordPreparedness Framework v2
  3. Next stopPetri
See it on the map

Same week

Gemma 3 QAT int4OfficialOpenAI launches o3 and o4-mini, with thinking with imagesOfficialCohere releases Embed v4OfficialOpenAI releases GPT-4.1 with a separate prompting guideOfficial

Nearby on the same axis

Reasoning Models Don't Always Say What They ThinkPaperOn the Biology of a Large Language ModelOfficialGrok inserts South African "white genocide" into unrelated answers for hoursPressAuditing Language Models for Hidden ObjectivesOfficial