Perplexity is an AI search tool, reviewed for consumer training defaults, enterprise exceptions, sharing, tracking, retention, and controls.
65
Mixed
65
Mixed
Perplexity can use searches and chat activity to improve AI systems, even though it offers an opt-out.
Perplexity is clearer on third-party sharing and consumer controls than on training, with visible rights and settings for account management.
Searches, prompts, and outputs can still be used for model improvement unless the user changes the setting, which is the main privacy risk.
50
Weak
Perplexity says some consumer data may be used to train AI models and improve search quality unless the user opts out.
Inputs, prompts, and linked-service activity can still feed AI-powered product functionality even when the service is not presented as a simple search box.
75
Good
65
Mixed
55
Weak
70
Mixed
Legal Privacy Policy
Open: perplexity.aiLatest Finding
Open: perplexity.aiTerms Of Service
Open: perplexity.aiHub Security
Open: perplexity.ai1. AI Use
Perplexity says some consumer data may be used to train AI models and improve search quality unless the user opts out.
2. AI Use
Inputs, prompts, and linked-service activity can still feed AI-powered product functionality even when the service is not presented as a simple search box.
3. Perplexity Enterprise
Perplexity says Enterprise Pro and API-style business contexts are governed separately from the ordinary consumer privacy policy.
4. AI Use
Perplexity says some users can opt out of data being used to train AI models through account settings.
5. AI Use
Perplexity says enterprise AI providers operate under strict zero-data-retention and zero-data-training agreements for enterprise data.
Founded
Unknown
Founder
Unknown
Parent Company
Perplexity
Lifecycle
Active
Category
AI Assistants & Models
CEO
Unknown
Security Team
In house
Date Added
05-18-2026
Once you delete a chat, you cannot recover it. Deleting a chat removes it both from your visible chat history and the system after the retention window.
Perplexity says Enterprise Pro and API-style business contexts are governed separately from the ordinary consumer privacy policy.
Perplexity says some users can opt out of data being used to train AI models through account settings.
Perplexity says enterprise AI providers operate under strict zero-data-retention and zero-data-training agreements for enterprise data.
75
Good
Perplexity says enterprise data is protected by stricter vendor agreements than ordinary consumer use.
The company describes a risk-based framework for reviewing subprocessors and vendors.
Some sharing paths, like social features and email/calendar connections, are optional user choices.
The privacy policy still allows sharing through vendors, business partners, linked email/calendar services, social features, and legal recipients.
65
Mixed
The policy says Perplexity and third parties collect usage data through cookies, pixels, tags, and other tracking technologies.
The policy also allows advertising and data-enrichment information to customize marketing.
Perplexity documents browser and cookie controls instead of treating tracking as untouchable.
The enterprise/business posture is less dependent on advertising than the ordinary consumer website layer.
The company is explicit about analytics and advertising information sources.
55
Weak
The privacy materials do not offer a simple, short, universal deletion window for prompts, usage logs, and linked-service data.
Logs, safety needs, legal obligations, and connected-service features can all extend how long data remains in the system.
Perplexity gives users a data request path and privacy rights contact channel.
Enterprise and API-style business data is governed under separate contractual arrangements instead of the consumer policy alone.
Account settings and opt-out choices can reduce some ongoing data use.
70
Mixed
Controls are spread across account settings, cookie behavior, and separate business-versus-consumer product lines.
Linked-service and partner features can still widen the data surface beyond the basic search experience.
Users can opt out of some model-training use in account settings.
Cookie and browser controls can limit some tracking behavior.
Optional social, email, and calendar connections are user-directed rather than mandatory.
