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 Terms of Service prohibit a range of uses of the Pi service including biometric categorization to infer protected characteristics, mass public surveillance, emotional recognition in workplace and educational settings, social scoring, and criminal profiling. These prohibitions are stated to apply to users of the service, not to Inflection AI's own processing.
This analysis describes what Inflection 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
These prohibitions reflect restrictions consistent with EU AI Act prohibited practices and high-risk AI system provisions, applied as user-facing acceptable use constraints. Violations of acceptable use terms are stated as grounds for enforcement action under the Terms of Service.
The updated policy establishes broader data collection practices than previously disclosed. The terms now explicitly state the company collects voice and audio inputs alongside text, whereas prior language specified only text and other materials. Additionally, the policy now discloses collection of precise geolocation information with user consent and authorization to access contacts, emails, calendars, and documents from third-party platforms. You can stop collection of precise location information at any time through the Your Choices section.
View change record →Under this clause, users are contractually prohibited from using Pi for biometric categorization, mass surveillance, workplace or educational emotional recognition, social scoring, and criminal profiling. Account enforcement or suspension may result from violations of these acceptable use terms as stated in the Terms of Service.
Cross-platform context
See how other platforms handle Acceptable Use Prohibitions Including EU AI Act-Aligned Restrictions and similar clauses.
Compare across platforms →Monitoring
Inflection AI has changed this document before.
Receive same-day alerts, structured change summaries, and monitoring for up to 20 platforms.
"Biometric Categorization: You may not use or attempt to use our Services to categorize people based on their biometric data to infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation. Surveillance or Facial Recognition: You may not use or attempt to use our Services to carry out mass public surveillance. Emotional recognition: You may not use or attempt to use our Services to infer people's emotions in the workplace or in educational settings except for medical or safety reasons. Social scoring: You may not use or attempt to use our Services for the purposes of evaluating or classifying the social behaviors of people to give them a social score or rating. Criminal Profiling: You may not use or attempt to use our Services for the purpose of profiling persons in order to assess or predict the risk of a person committing a criminal offence.Excerpt from Inflection AI's Privacy Policy
1. REGULATORY LANDSCAPE: These prohibitions are substantively aligned with provisions of the EU AI Act addressing prohibited AI practices, including real-time remote biometric identification in public spaces, social scoring, and emotional recognition in regulated contexts. The EU AI Act applies to providers and deployers of AI systems in the EU; these terms impose analogous restrictions on users as a contractual matter. The FTC Act is engaged to the extent these restrictions reflect deceptive or unfair practice considerations. Relevant enforcement authorities include EU AI Act national market surveillance authorities and the FTC. 2. GOVERNANCE EXPOSURE: Medium. The prohibitions are user-facing contractual restrictions, not direct regulatory compliance statements by Inflection AI. Enterprise and API customers deploying Pi in business contexts should assess whether their use cases are consistent with these restrictions, particularly in HR, education, law enforcement, or public safety contexts. 3. JURISDICTION FLAGS: EU and EEA deployers face the most significant exposure because the EU AI Act creates direct regulatory obligations for providers and deployers of AI systems in these categories. U.S. enterprise customers should assess whether proposed use cases implicate the prohibited categories, as violations would constitute a breach of the Terms of Service. 4. CONTRACT AND VENDOR IMPLICATIONS: Organizations procuring Pi for enterprise use should review their intended use cases against each prohibited category before deployment. The prohibition on using outputs to develop models that compete with Inflection AI, stated elsewhere in the Acceptable Use section, is a separate commercial restriction relevant to API customers and technology vendors. 5. COMPLIANCE CONSIDERATIONS: Legal teams advising enterprise customers should evaluate whether proposed deployments of Pi in HR, education, law enforcement, or security contexts fall within the prohibited categories. Compliance with these restrictions should be documented as part of vendor onboarding and ongoing acceptable use monitoring. The prohibition on reverse engineering and competitive model development is separately relevant for technology companies and should be flagged in procurement and legal review.
These prohibitions reflect restrictions consistent with EU AI Act prohibited practices and high-risk AI system provisions, applied as user-facing acceptable use constraints. Violations of acceptable use terms are stated as grounds for enforcement action under the Terms of Service.
Under this clause, users are contractually prohibited from using Pi for biometric categorization, mass surveillance, workplace or educational emotional recognition, social scoring, and criminal profiling. Account enforcement or suspension may result from violations of these acceptable use terms as stated in the Terms of Service.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Inflection AI.