The evidence behind the guides

Sources

Find a study, check what it supports, and follow it back to a concern.

TopicCollection

2 sources

Guide citations are still under editorial review.

Qi and colleagues Research study

Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To! (original source, new tab)

Tests safety changes under both malicious and benign customization; accidental weakening is distinct from intentional poisoning.

Mitchell and colleagues · FAT* 2019 Research proposal

Model Cards for Model Reporting (original source, new tab)

Proposes documentation of intended use, evaluation conditions and performance; a card's existence is not proof of quality.

Sources retain their original publishers’ rights. Inclusion does not imply endorsement. Saved sources stay in this browser on this device. Dates and source descriptions come from the existing editorial records.