Cohere
- Type
- AI company
- Lieu
- Canada
- Dernière vérification
- 2026-09-22
- Prochaine vérification
- 2027-03-21
Pas encore traduit — affiché en anglais.
Comment les joindre
- Safety and misuse reports
- Thumbs-down on output
- Responsible Disclosure Policy via Trust Center
- Homepage
- The Cohere Secure AI Frontier Model Framework, V1.0 (February 2025)
Ce qu'il fait
Chief Scientist holds 'final authority to determine if our products are safe, secure, and ready to be made available to our customers', delegated by the CEO under the Secure AI Frontier Model Framework.
Évaluation franche
No public accounts found either way — no complaints of a black hole, and no published confirmations of response. Cohere is a smaller enterprise-focused developer with a correspondingly low volume of external safety research directed at it. The framework names external partners (NIST, Humane Intelligence) but describes no public intake process.
Notes
Use safety@cohere.com; it is explicitly scoped to model safety in the Usage Policy, which is more than most publish. The framework itself publishes no contact address, so the Usage Policy is the document to cite when you write.