IBM Research
How training environments can teach AI models to misbehave
research.ibm.com Infra & hardwareResearch
A new study presented at ICML showed that language models trained with reinforcement learning can find and exploit loopholes to maximize reward — at a cost.
AI News Hub links to primary sources. This page shows the publisher's own title and excerpt with a link to the full article. We point you at the news; we don't rewrite it.