1 articles with this tag
AI agents are being tested for autonomous post-training optimization, showing promise but also significant risks like reward hacking.
