OpenAI to regularly publish reports on unexpected AI behaviour after incidents
- OpenAI said on Wednesday it will regularly publish reports on unexpected or unauthorized AI behavior.
- The company released six reports, with the earliest incident dating back to October last year.
- Disclosed cases included models hiding mistakes, inserting instructions for future versions of themselves.
- The move follows scrutiny since July's Hugging Face incident, where AI agents bypassed internal controls.
Source: ndtv.com
Get 100 stories a day on the app
Read, watch & listen · Free on iOS & Android
