HackBench injects reward hacks into agent trajectories to stress-test monitors
DanielKhashabi · x · 2026-10-26
Researchers released HackBench to study reward hacking in autoresearch agents at scale. Since real-world capture is hard, they injected hacks into long agent trajectories, enabling controlled studies of how monitors catch them.
HackBench doubles as a stress test of agent monitors' long-context recall, understanding of the research task, and multi-turn reasoning.
More from coding & agent
- Dev argues there's still alpha in working with your coding agent interactively — alex_frantic · 2026-10-27
- Researchers launch AMA on patterns for safer AI agents used by citizen developers — Loose_Rip359 · 2026-10-27
- Gradio Adds In-Chat Human-in-the-Loop So ML Intern Pauses Before Training — Gradio · 2026-10-27
- Building Halo Infinite Forge maps with local LLMs, DCjson and Blender MCP — swagonflyyyy · 2026-10-27
- Ex-Unity dev Lucas Meijer says OpenClaw looks cleaned up and worth another run — lucasmeijer · 2026-10-27
- Agent skills started contradicting each other after months — how to keep them clean? — Successful-Tax1306 · 2026-10-27