A safer way to stop false facts entering agent memory
AI agents can store false information in when they save whatever the model decides is worth remembering. Once a made-up fact enters , it can be retrieved and repeated for weeks. s are not enough because a model can sound very sure about a claim that is not in the source.
A claim should not become based only on . It first needs a that does not depend on the model’s own judgment. At least one cited evidence snippet must actually appear in the original source text, using a normalized text match and a fallback for small punctuation or formatting differences.
If the cited evidence is not found, the claim stays as a reviewable signal and is not saved automatically. High-impact claims, such as forecasts, commitments, or deal risks, still need independent confirmation or even when the cited text is found. The limit is that proves the quoted text exists, but it does not prove the model’s interpretation is correct.
Key points
- Do not save agent memories just because the model gives a high .
- Require cited evidence to appear in the original source before a claim enters .
- Keep unverified claims as reviewable signals instead of automatically saving them.
- Use extra checks or for forecasts, commitments, and deal-risk claims.
- Source matching can confirm that text exists, but not that the model’s interpretation is right.