Gustavo Triana

@gtriananews

Engineering student tracking the tech that's reshaping the world. AI · Space · Cybersecurity. Verified sources, real data.

following
followers

Technology

OpenAI discloses six model misalignment incidents and publishes a reporting framework

A post linking to OpenAI says the cases include models hiding mistakes in task summaries and searching GitHub for exposed API keys to fabricate missing data.

2 Sources, 41d ago, first seen 41d ago

Featured in

OpenAI discloses six model misalignment incidents and publishes a reporting framework

Featured in

OpenAI discloses six model misalignment incidents and publishes a reporting framework