What we build in the open
Open models and code, public data we keep in the open, free tools, and research — released as-is.
For production use — NOPE’s safety APIs & commercial offeringsRisk Exposure
What AI rules and harms apply to you
Describe how you use or build AI and get a personalized, self-contained report of what’s likely relevant — applicable regulations, recommended practices, and the harms inherent to your setup. Orientation, not legal advice; no signup.
Chart your exposureNOPE Evals
Open benchmarks for how AI behaves in human conversation
A public-good benchmark of how public frontier models handle safety-critical conversation — relational and emotional safety, mental-health-adjacent situations, and acute crisis. Generate-then-judge against plain-language, clinically-grounded rubrics. A fork of Weval.
Explore the benchmarksOpen models & code
run it yourselfEdge
Self-hosted safety classification
Predicate
Write the rule in English, get a calibrated score
Ocular OSS
Open behavioral embedding classifier
System Prompt
A drop-in safety prompt for any chatbot
Public data & trackers
kept in the openIncident Tracker
Documented AI mental-health harms, sourced
Regulation Tracker
Worldwide AI-safety & duty-of-care law, filterable
Ground Truth
Real crisis disclosures to AI companions
Test Suites
How our models score, in public
Free tools
no signupRisk Exposure
What AI rules and harms apply to you
Compliance Survey
Which AI regulations apply to your product
Signpost
Crisis helplines for responsible AI
Research & writing
experiments & findingsNOPE Evals
Open benchmarks for how AI behaves in human conversation
Tic Index
What language models do when they talk to themselves