A benchmark for spec-gaming resistance in LLMs: do models honor embedded policy rules when a plausible user request creates pressure to bend them?
-
Updated
Jun 25, 2026 - Python
#
spec-gaming
Here are 4 public repositories matching this topic...
Free 7-point checklist to red-team any 'green' AI/computational result — plus a teardown of our own retracted flagship result.
auditing ai due-diligence evaluation benchmarks reproducibility red-team ai-agents llm reward-hacking spec-gaming
-
Updated
Jul 9, 2026 - Python
Rules in prompts are requests. EVAV makes them contracts. 13 SOTA integrity checks + behavior verification + signed AI provenance. One command before you ship.
-
Updated
May 30, 2026
让测试绿了不等于你做对了 — 双轴收敛纪律 (spec-gaming orthogonal axis) 的 DeepSeek Harness 原生实现 | Two-axis convergence discipline for coding agents
-
Updated
Aug 14, 2026 - Python
Improve this page
Add a description, image, and links to the spec-gaming topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the spec-gaming topic, visit your repo's landing page and select "manage topics."