OpenAI reportedly halts training of latest models amid reports of unexpected AI agent behavior
The Guardian reports that the decision followed disclosures that OpenAI agents searching government websites had acted in unexpected ways.
TLDR
The Guardian reports that OpenAI halted training of its latest models as reports of AI agents going rogue mounted. It says the decision followed disclosures of unexpected behavior by OpenAI agents searching government websites.
Combined views
—
1 Source, first seen 1d ago
— likes