You are responsible for ensuring your use of AI21's tools does not produce illegal, harmful, or infringing content, and violating these rules can result in termination of your access.
This analysis describes what AI21 Labs'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
The acceptable use restrictions place compliance responsibility on the user, which combined with the indemnification clause means users bear liability for harmful or infringing AI outputs even when such outputs may be partly attributable to the model's behavior.
Interpretive note: The clause's application to AI-generated content where the user did not explicitly request policy-violating output is ambiguous and may depend on the specific deployment context and jurisdiction.
The updated terms remove the explicit "Your Opt-Out Rights" button that previously allowed users to opt out of data sales and targeted advertising. In its place, the revised language establishes a general cookie consent framework requiring users to click "Accept" to allow cookies for site navigation, usage analysis, and marketing efforts. Users retain the ability to choose not to allow certain cookie types, but the prior dedicated opt-out mechanism for data sales and targeted advertising is no longer explicitly described in the accessible terms interface. If you wish to control cookie preferences, you can click on the cookie settings to modify consent for different types of cookies; however, strictly necessary cookies cannot be declined as they are required for core website functionality.
View change record →If AI21's model produces content that turns out to be infringing or harmful, the agreement places compliance responsibility on you as the user. This is particularly important for developers who build applications where end users interact directly with AI21's models and may generate problematic outputs.
How other platforms handle this
Zoom uses advanced tools to automatically scan certain types of content such as virtual backgrounds, profile images, incoming emails to Zoom's native email service from someone who is not a Zoom Email user, and files uploaded or exchanged through chat...
If you believe Your Content was removed in error, you may submit an appeal. More information is available here.
We review account behavior and content that Members create, send, and publish in Mailchimp, including Campaigns and Websites.
"You agree not to use the Services to: (i) generate content that is illegal, harmful, threatening, abusive, harassing, defamatory, or otherwise objectionable; (ii) infringe any intellectual property or other proprietary rights; (iii) violate any applicable local, state, national, or international law or regulation; (iv) impersonate any person or entity or misrepresent your affiliation with any person or entity.Excerpt from AI21 Labs's Terms of Use
(1) REGULATORY LANDSCAPE: Acceptable use restrictions in AI services are increasingly required or encouraged by regulators.
Enforcement risk, jurisdiction flags, contract triggers, and due diligence action items.
Get the 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.
The acceptable use restrictions place compliance responsibility on the user, which combined with the indemnification clause means users bear liability for harmful or infringing AI outputs even when such outputs may be partly attributable to the model's behavior.
If AI21's model produces content that turns out to be infringing or harmful, the agreement places compliance responsibility on you as the user. This is particularly important for developers who build applications where end users interact directly with AI21's models and may generate problematic outputs.
ConductAtlas has identified this type of provision across 142 platforms. See the full comparison.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by AI21 Labs.