The reasoning was protected. The interaction boundary was less complete.
AI-generated conceptual illustration; not incident footage.
Executive summary
OpenAI says a July campaign manipulated model interactions to expose protected reasoning. It reports 16,000 attempted extraction-pattern requests on July 24–25. September 30 is the disclosure date.
Context: Company disclosure of human-directed activity, not an autonomous escape. Attempt counts are not successful extractions. OpenAI says encryption, databases and stored user conversations were not compromised; protections were strengthened. Related research predates September.
Source published 2026-09-30 · Selected claims checked 2026-10-01
July 1–28 activity; disclosed September 30
Published 2026-10-01T15:14:32.212422Z
What happened
OpenAI reports a coordinated July campaign against protected reasoning and describes mitigations to a replay pathway. Its request counts describe attempts, not necessarily successful extractions. Related independent researchers describe cross-session and cross-model reasoning exposure in an August 10 preprint; that paper is background rather than a new experiment today.
The 2050 museum label
The secret stayed encrypted. The interface did the talking.
Company disclosure of human-directed activity, not an autonomous escape. Attempt counts are not successful extractions. OpenAI says encryption, databases and stored user conversations were not compromised; protections were strengthened. Related research predates September.
Original reporting, primary documents and public posts. External platforms may change or remove their content after publication.
September 30 company disclosure; reported counts describe attempts, not necessarily successful extractions
Related August 10 preprint; abstract and submission history reviewed; no underlying exposed data retrieved
Follow the museum
Dispatches from 2050
We’ll email you new exhibits, with the story, the sources and the museum’s commentary.