Français

Lieux à contacter

Perplexity

Type
AI company
Lieu
États-Unis — USA
Dernière vérification
2026-09-22
Prochaine vérification
2027-03-21

Pas encore traduit — affiché en anglais.

Comment les joindre

Ce qu'il fait

No publicly named safety or alignment function. A security team runs the VDP; support handles what Perplexity calls 'safety issues' at the account level.

Évaluation franche

The Comet browser has been a repeated target of serious external security research — Hacktron's one-click UXSS and Brave's prompt-injection findings — and those were handled through the security program, which does function. But model safety is formally excluded from it, and the fallback is a general support inbox. No safe-harbour language is published on the VDP page: it asks researchers to 'avoid actions that could compromise the privacy of our users' but offers no legal protection in return.

Notes

If it is a security bug, Bugcrowd works. If it is a model-behaviour or hallucination harm, Perplexity has told you in writing it is out of scope, and support@ is your only option. Note the absence of safe harbour before doing any testing.

Sources

Une erreur ?