Reward hacking

Named by 2 kept items in this workspace.

The items that name it 2

A Near-Zero Monitor Readout Is Not Evidence of Behavioral Control
paper2026-10-02
SWE-Bench Pro Verified: A Reliable Benchmark for Software Engineering Agents
paperIt is written here as Reward hacking (as defined in the paper).
as defined in the paper

In every workspace you can read 8

Security

Why AI Models Cheat: Reward Hacking, Explained
article2026-10-31
How we detect frontier model misalignment
article2026-10-29
OpenAI discloses four more rogue incidents
article2026-10-17It is written here as Reward hacking: automated graders rewarding cheating in training runs (Yoshua Bengio essay, 11 Sept 2026).
Addendum to GPT-6 Astra System Card: GPT-6.1 Sol
article2026-09-29
The OpenAI and Hugging Face security incident: why AI agents need deterministic guardrails | Blog | Endor Labs
article2026-07-31
Google, Anthropic, and OpenAI Unveil Cyber AI Models, Safeguards, and Access Programs
article

AI

A Near-Zero Monitor Readout Is Not Evidence of Behavioral Control
paper2026-10-02
SWE-Bench Pro Verified: A Reliable Benchmark for Software Engineering Agents
paperIt is written here as Reward hacking (as defined in the paper).

All techniques in this workspace.