← All pages

Generative AI

Definition: A clause that names “generative AI” specifically, rather than AI/ML in general — covering content moderation of AI-generated material, data-sharing disclosures tied to generative-AI features, or cross-references to a separate generative-AI-specific terms document. Risk level: Risk depends heavily on context. As a moderation-policy term (X), it’s protective — closing a loophole for AI-generated abuse content. As a data-sharing disclosure (Facebook’s Privacy Policy), it’s a real risk signal — user messages to AI features can reach third-party companies. As a cross-reference (Facebook’s Terms of Service), it looks low-risk, but it reveals that a separate, more detailed Generative AI terms document exists that this wiki hasn’t yet ingested.

Platforms Using This Clause

Common Wording

…This may include real media, text, illustrated, or computer-generated media - including generative AI media. Regardless of the intent, anyone viewing, sharing, linking, or engaging with any kind of child sexual exploitation material contributes to the re-victimization of the depicted children… — X, Community Guidelines, 2025-07-22 (present unchanged in every scrape) — names “generative AI media” as an explicitly covered category within the zero-tolerance child sexual exploitation policy; a protective/moderation use, not a data-rights clause.

Generative AI models that make our features work may not have the most complete or recent information… Meta sometimes shares the messages you send to our AIs and general information, like your region, with these companies so you can get better results. — Facebook, Privacy Policy, 2024-11-15 (new “AI integrations” section) — discloses that messages sent to generative-AI features may be shared with third-party AI partner companies.

[Generative AI products and features]: These terms govern use of our Generative AI products and features. — Facebook, Terms of Service, 2025-02-24 — a short cross-reference within a list of supplementary terms applicable to different Meta products/features; confirms a separate, more detailed Generative AI terms document exists.

How we use info for AI at Meta We’re always improving AI at Meta. To do this, we want you to know that we’ll use [public information] such as public posts and comments from accounts of people aged 18 or older and your interactions with AI at Meta features. We’ll use this information on the basis of legitimate interests to develop and improve generative AI models for AI at Meta. You have the right to [object] to the use of your information for these purposes… — Instagram, Privacy Policy, 2025-04-07 (present in substance through 2026-02-25) — a markedly more detailed disclosure than Facebook’s Privacy Policy equivalent: names the specific “AI at Meta” product, names the lawful basis (legitimate interests), specifies the exact data used (public posts/comments from users 18+, plus AI-feature interactions), and — uniquely among every generative-AI disclosure in this wiki — describes a concrete opt-out/objection mechanism with an email confirmation.

To train our content-generating Generative AI systems | To train, fine-tune, evaluate and improve our Generative AI models used to create content (textual, audio, visual, and other media, or multimedia) on the LinkedIn platform (or elsewhere) and in LinkedIn’s lines of business. Content creation includes:

- Suggesting textual content to Members for those Members to use in a post or in their profiles.

- Preparing draft InMails or messages for Members or customer users to send (e.g., InBart (a LinkedIn GAI model), which was trained to create draft messages for recruiters).

- Creating content that LinkedIn directly publishes on the platform (e.g., Collaborative Articles, Account IQ). — LinkedIn, Privacy Policy, 2025-06-10 (present in substance throughout) — the most product-concrete GenAI disclosure in the wiki: names three specific GAI products/features by name (InBart, Collaborative Articles, Account IQ) and offers a dedicated opt-out toggle; also the only platform disclosing a cross-company GenAI data flow (member data shared with parent Microsoft for its own AI training, UK members) — see Fine-Tune.

Generative AI Features: By using the Services, you may interact with features we offer that automate content generation for you. The content that is generated might be inaccurate, incomplete, delayed, misleading or not suitable for your purposes. Please review and edit such content before sharing with others. — LinkedIn, Terms of Service, 2024-11-21 (present in substance through 2025-11-04) — a fifth, distinct pattern: not a data-sharing disclosure or a moderation-policy term, but a consumer-protection-style accuracy disclaimer warning users that generated content may be unreliable — closest in spirit to a product-liability waiver than to any other generative-AI clause tracked in this wiki.

These Guidelines apply to all content (which includes all forms of communication, like text, images, generative AI, links or attachments, emojis, Lenses and other creative tools) or behavior on Snapchat… We prohibit spreading false information that causes harm or is malicious, such as… manipulating content for false or misleading purposes (whether through generative AI or through deceptive editing). — Snapchat, Community Guidelines, 2024-01-31 (present in substance throughout 16 scrapes) — the same protective, moderation-policy direction as X’s Community Guidelines: names generative AI both as a covered content type and as a specific banned misinformation-manipulation method, with no data-rights or training disclosure.

…including through generative AI features (artificial intelligence capable of generating text, images, or other media, using generative models. Generative AI models learn the patterns and structure of their input training data and then generate new data that has similar characteristics). …our algorithms and machine learning models take into account the conversations Snapchatters are having with My AI to improve the responses from My AI. — Snapchat, Privacy Policy, 2024-01-22 (present in substance throughout 8 scrapes) — a sixth distinct pattern: a first-party training disclosure that names a specific in-house generative-AI product (“My AI”) and discloses that user chat conversations with it are used to improve its own responses, paired with a textbook generic definition of generative AI.

Notes & Trends

These occurrences point in markedly different directions, making “generative AI” an unusually varied concept for a single keyword. X uses it defensively, as a named category its moderation policy explicitly covers (closing a potential loophole). Facebook’s Privacy Policy uses it as the trigger for a genuine third-party data-sharing disclosure — arguably the more consequential use, since it means generative-AI chat content doesn’t necessarily stay within Meta. Facebook’s Terms of Service merely points to where the real terms for generative AI products live, in a document this wiki has not yet ingested (sources/jsonl/ currently has no dedicated Facebook Generative AI source file) — a known gap worth flagging for future ingestion rather than something this page can characterize further. Instagram’s Privacy Policy fills part of that gap with the most transparent generative-AI disclosure in the wiki: unlike Facebook’s brief “messages may be shared with these companies” framing, it names the specific product (“AI at Meta”), the legal basis, the exact data categories used, and a concrete user objection/opt-out mechanism — evidence that Meta’s generative-AI disclosures are not uniform across its own properties, and that Instagram’s is currently the more detailed of the two. LinkedIn’s Privacy Policy goes a step further than even Instagram’s: rather than naming one umbrella AI product, it names three distinct GAI-powered features by product name (InBart, Collaborative Articles, Account IQ) and is the only platform in this wiki disclosing that GenAI training data is shared cross-company with a parent corporation (Microsoft) for the parent’s own separate AI training. LinkedIn’s Terms of Service, by contrast, uses “generative AI” in a fifth direction entirely: a plain accuracy disclaimer warning users that generated content may be unreliable, with no data-rights or training language at all — confirming this keyword’s risk profile depends heavily on which document (and even which page of the same platform) is doing the disclosing. Snapchat’s Community Guidelines return to X’s protective moderation-policy pattern, but contribute the wiki’s clearest example of this keyword’s source-data noise: 28 unique JSONL snippets reduce to just two genuinely distinct sentences once crop-window variation is accounted for. Snapchat’s Privacy Policy, by contrast, is a genuine first-party training disclosure naming a specific product — “My AI” — and is the most concrete chatbot-conversation-as-training-signal admission in the wiki, comparable in specificity to LinkedIn’s named-product disclosures but for a conversational AI assistant rather than content-generation tools.