Chiron stamps AI outputs only when it can prove them exactly
justkidding1908 · reddit · 2026-07-23
Chiron only stamps AI outputs when it can prove them exactly
Chiron is presented as an exact-or-refuse evidence gate for structured outputs and supported claims.
- For sequence-like data, it infers a constrained rule and tests it on held-out terms; only exact prediction becomes verified: true.
- For text claims, it marks each claim as VERIFIED, REFUTED, or REFUSED and shows the coverage boundary instead of silently approving the whole response.
- The author publishes a bounded external eval with 22 stamped outputs, 22 correct, 0 false stamps, and 12 refusals, and also discloses an earlier failure set that was fixed and rerun.
- The repo includes grading and challenge scripts, and is source-available for noncommercial use with commercial licensing for deployment.
More from Research
- Robotics paper says VLA and world models are not enough for grounded supervision — hbouammar · 2026-07-23
- AI could compress decades of biomedical research into days, says Derya Unutmaz — DeryaTR_ · 2026-07-23
- OpenAI and Apollo show models may optimize graders, not user intent — rohanpaul_ai · 2026-07-23
- AI Autonomously Disproves Decades-Old Math Conjectures: The Singularity's Opening Phase — imjustnewatai · 2026-07-23
- Applied Math Dominates AI, But Why Does Gradient Descent Actually Work? — fkasummer · 2026-07-23
- Cursor’s Composer 2.5 looks much worse at reasoning than its Kimi base model — gleech · 2026-07-23