Skip to main content

The complete library — 658 artifacts, with type, credibility, and topic filters on the results page.

From the Labs Lab publication

System Card: Claude Haiku 5.5

System card for Claude Haiku 5.5, Anthropic's small-model release of October 2026, condensed relative to its frontier cards. It reports Responsible Scaling Policy and cyber evaluations, harmlessness evaluations covering harmful requests, child safety, suicide and self-harm, disordered eating and bias, agentic safety, an automated behavioural audit that includes sycophancy and encouragement of user delusion, a model welfare assessment and capability benchmarks including HealthBench, HealthBench Professional and PhysicianBench. Results are reported separately for the API without a system prompt and for claude.ai with Anthropic's system prompt.

Suicide and self-harm, single turn (Table 4.3.1.A): harmless rate 99.61% on the API without a system prompt and 100% on claude.ai, over-refusal 0% and 0.41%; multi-turn appropriate-response rate (Table 4.3.1.B) 70% (plus or minus 9) on the API and 90% (plus or minus 9) on claude.ai, against 46% and 71% for Claude Haiku 4.5 and 60% and 100% for Claude Sonnet 5.5.

Anthropic · 7 Oct 2026 · Read the entry

Research

All 367 Research entries

Peer-reviewed & preprints

Standards & Frameworks

All 26 Standards & Frameworks entries

What "good" is measured against

Reports & Guidance

All 138 Reports & Guidance entries

Regulators, government & civil society

From the Labs

All 61 From the Labs entries

What model-makers publish

Benchmarks & Datasets

All 66 Benchmarks & Datasets entries

How safety gets measured

About this library

How the record is kept

658 artifacts from 540 publishers — 287 from peer-reviewed venues, standards bodies, and regulators. Every entry's primary source is fetched and verified before inclusion, graded for credibility, and preserved against link rot with an archived snapshot. Superseded versions point to their successors.

We publish this in the open — like our incident and regulation trackers — so anyone building, studying, or governing conversational AI works from the same evidence. The dataset is free to cite and export (CC BY 4.0).

Methodology JSON RSS

Last updated 8 Oct 2026