Get the weekly research letter
Companies change their terms quietly. We read every version and catch what actually changed. One email a week on the changes that matter and what they mean. No account.
The policy states that user queries and received responses may be used to train and improve Perplexity's AI models, with an opt-out available through privacy settings or by emailing privacy@perplexity.ai. The default position is that this use applies unless the user actively opts out.
This analysis describes what Perplexity AI's agreement states, permits, or reserves. It does not constitute a legal determination about enforceability. Regulatory applicability and practical outcomes may vary by jurisdiction, enforcement context, and individual circumstances. Read our methodology
This provision establishes an opt-out default for use of personal interaction data in AI model training, meaning training use proceeds unless users take affirmative action. For EU/EEA users, the adequacy of an opt-out mechanism as opposed to opt-in consent for this processing purpose may require evaluation against GDPR requirements depending on the processing basis asserted.
Interpretive note: The lawful basis asserted under GDPR for AI training data use is not explicitly named in the reviewed policy text, creating uncertainty about whether the opt-out mechanism is sufficient for EU/EEA users.
The updated Privacy Notice establishes more granular disclosure of data collection methods across multiple product areas. Perplexity now explicitly discloses that it collects and stores browsing history and settings in the Comet browser based on consent or legitimate interest, accesses email content through Email Assistant to analyze messages (while stating it does not train AI models on that content), and collects demographic data if users voluntarily upload it. The revised structure also clarifies that local browser data storage occurs on users' devices and that incognito mode does not fully prevent tracking by websites or Perplexity. You can review Comet privacy settings and controls as described in the updated policy.
View change record →Severity upgraded from medium to high, and new opt-out mechanism added with specific contact method and privacy settings option.
View full change record →Under this clause, query content and conversation responses are used for AI model training by default, and users must actively opt out to prevent this use. The opt-out is available via platform privacy settings or by contacting privacy@perplexity.ai.
How other platforms handle this
You can limit to what extent we use your personal information for these purposes.
In certain circumstances, the right to data portability, which means that you can request that we provide certain Personal Data we hold about you in a machine-readable format
For data portability requests, We will select a format to provide Your personal information that is readily useable and should allow You to transmit the information from one entity to another entity without hindrance.
Monitoring
Perplexity AI has changed this document before.
Receive same-day alerts, structured change summaries, and monitoring for up to 20 platforms.
"We may use your interactions with our services, including your queries and the responses you receive, to train, fine-tune, and improve our AI models and services. You may opt out of this use of your data by adjusting your privacy settings or contacting us at privacy@perplexity.ai.Excerpt from Perplexity AI's Privacy Policy
1) REGULATORY LANDSCAPE: This provision implicates GDPR provisions on lawful bases for processing, purpose limitation, and data subject rights, as well as CCPA provisions on use of personal information beyond the disclosed purpose. The UK ICO and EU supervisory authorities have issued guidance on the use of personal data for AI training. The FTC has signaled scrutiny of AI training data practices under unfair and deceptive practices standards. 2) GOVERNANCE EXPOSURE: High. The opt-out default for AI training data use is the most operationally significant provision in the policy for EU/EEA compliance. GDPR generally requires a valid lawful basis for each processing purpose; if legitimate interests is the asserted basis, a balancing test must be documented. If consent is the basis, opt-out is not a sufficient mechanism under GDPR and opt-in would be required. 3) JURISDICTION FLAGS: EU/EEA and UK users face heightened exposure due to GDPR and UK GDPR requirements on lawful basis and consent. California users have the right under CPRA to opt out of use of sensitive personal information beyond what is necessary to provide services. Illinois and other states with emerging AI transparency laws may impose additional disclosure requirements. 4) CONTRACT AND VENDOR IMPLICATIONS: Enterprise or API customers should assess whether their use agreements with Perplexity include provisions restricting the use of their users' data for model training, as the default policy permits this use absent opt-out. B2B contracts may need to address this use right explicitly. 5) COMPLIANCE CONSIDERATIONS: Legal teams should document the lawful basis asserted for AI training data use by jurisdiction, ensure the opt-out mechanism is technically functional and accessible, and assess whether supplementary consent notices are required for EU/EEA users. Data mapping should classify training data use as a distinct processing purpose with its own retention and access controls.
Regulatory citations, enforcement risk, and due diligence action items.
Ad personalization controls removed. Contact scanning added. Advertiser data partnerships quietly dropped. A timeline of every change.
Provision-level monitoring, governance timelines, and regulatory mapping built from archived source documents and historical version tracking.
This provision establishes an opt-out default for use of personal interaction data in AI model training, meaning training use proceeds unless users take affirmative action. For EU/EEA users, the adequacy of an opt-out mechanism as opposed to opt-in consent for this processing purpose may require evaluation against GDPR requirements depending on the processing basis asserted.
Under this clause, query content and conversation responses are used for AI model training by default, and users must actively opt out to prevent this use. The opt-out is available via platform privacy settings or by contacting privacy@perplexity.ai.
ConductAtlas has identified this type of provision across 295 platforms. See the full comparison.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Perplexity AI.