The Instagram (Meta) record

AI-content labeling

• Meta began adding AI-content labels across a wide range of photo, video, and audio content in May 2024, applying them when it detects industry-standard AI indicators or when a creator discloses that an upload is AI-generated. In the platform's own words: “We began adding these labels when we detected industry standard AI image indicators or when people disclosed that they were uploading AI-generated content.”
  Source: Meta Transparency Center — Labeling AI Content · observed 2026-09-03
• Meta describes its own labeling methodology as still evolving and incomplete, cautioning that its reported figures are a snapshot and may not capture every piece of AI-edited content. In the platform's own words: “Our labeling methodology is still evolving. The data reported here represents a snapshot in time, and may not capture labels on some content that is edited with AI.”
  Source: Meta Transparency Center — Labeling AI Content · observed 2026-09-03
• How visible the label is depends on how much AI was involved: content only lightly edited with AI tools carries its label inside the post's three-dot menu, while content that is substantially AI-generated keeps the label visible on the post itself, per a Meta newsroom post dated September 2024.
  Source: About Meta newsroom — Meta's Approach to Labeling AI-Generated Content and Manipulated Media · observed 2026-09-03
• Meta does not remove undisclosed AI or manipulated content simply for being unlabeled. Its stated mechanism is reduced distribution once independent fact-checkers weigh in, with removal reserved for content that separately breaks another rule, regardless of whether AI was involved. In the platform's own words: “When fact-checkers rate content as False or Altered, we show it lower in Feed so fewer people see it.”
  Source: About Meta newsroom — Meta's Approach to Labeling AI-Generated Content and Manipulated Media · observed 2026-09-03
• Content made with Meta's own AI image generator carries a distinct Imagined with AI label, separate from the general AI info tag applied to AI content uploaded by others.
  Source: About Meta newsroom — Meta's Approach to Labeling AI-Generated Content and Manipulated Media · observed 2026-09-03
• For advertising, Meta requires disclosure whenever an ad image is created or significantly edited with generative AI — Meta's own tools or a third party's — though minor touch-ups like resizing or color correction don't trigger it. In the platform's own words: “Ad images created or significantly edited using our generative AI creative features or ad images created or edited using third-party AI tools.”
  Source: Meta Help Center — How AI-generated images in ads are identified and labeled on Meta · observed 2026-09-03
• Political, election, and social-issue ads face a stricter standard: Meta already requires advertisers to disclose AI use in those categories, independent of the general ad-labeling threshold. In the platform's own words: “For ads about social issues, elections or politics, advertisers are already required to disclose if the image, video or audio are created or edited with AI.”
  Source: Meta Help Center — How AI-generated images in ads are identified and labeled on Meta · observed 2026-09-03

Impersonation and likeness

This is Meta's general Community Standards rule against impersonation — it is not written specifically for AI content. A synthetic image used to impersonate a real person would be assessed under this policy together with the AI-labeling rules above.

• Meta's Community Standards ban creating or using accounts to deceive others by impersonating a real person's image, name, or likeness. In the platform's own words: “Impersonate another person or entity by: Using their image(s), name, or likeness with aim to deceive others.”
  Source: Meta Transparency Center — Authentic Identity Representation policy · observed 2026-09-03
• The rule reaches beyond direct impersonation to any identity misrepresentation used to mislead people, deceive them, or dodge enforcement. In the platform's own words: “Engage in identity misrepresentation to mislead or deceive others, evade enforcement, or violate Community Standards.”
  Source: Meta Transparency Center — Authentic Identity Representation policy · observed 2026-09-03
• Fan and parody accounts are permitted, but Meta requires them to clearly disclose their unofficial status, at minimum in the account name. In the platform's own words: “make its unofficial status abundantly clear at minimum in account name and ideally elsewhere on profile.”
  Source: Meta Transparency Center — Authentic Identity Representation policy · observed 2026-09-03

Verification

• A verified badge is available to subscribers of Meta's paid Meta Verified / Meta One plans, once Meta confirms the subscriber's identity on the subscribed profile. In the platform's own words: “Subscribers of the Meta One Essential, Meta One Advanced, Meta One Expert and Meta One Max plans are eligible for a verified badge after your identity is verified on your subscribed profiles.”
  Source: Meta Help Center — About the verified badge for Meta One subscribers · observed 2026-09-03
• Verification requires a government-issued ID, and in some regions a selfie video as well. In the platform's own words: “You may be asked to provide a government-issued ID and, in selected regions, may be asked to submit a selfie video.”
  Source: meta.com/meta-verified product page · observed 2026-09-03
• Once verified, subscribers get proactive impersonation monitoring as a bundled benefit — Meta says it will remove accounts it determines are pretending to be them. In the platform's own words: “Protect your brand with proactive impersonation monitoring. Meta will remove accounts that we determine are pretending to be you.”
  Source: meta.com/meta-verified product page · observed 2026-09-03
• The confirmed Standard-tier price is $14.99 per month, per profile. Meta's pricing for higher tiers is reported in press coverage but was not independently confirmed against a Meta-owned pricing page in this review, so it is left off this record. In the platform's own words: “$14.99/month per profile.”
  Source: meta.com/meta-verified product page · observed 2026-09-03

Enforcement

• Meta has not published a standalone penalty schedule for synthetic media. Its stated enforcement mechanic is labeling plus reduced distribution for content fact-checkers rate as false or altered, with removal reserved for posts that separately violate another Community Standard — voter interference, harassment, or violence, for example — independent of whether AI was involved.
  Source: About Meta newsroom — Meta's Approach to Labeling AI-Generated Content and Manipulated Media · observed 2026-09-03
• For impersonation specifically, Meta's stated enforcement range runs from temporary restriction to permanently disabling the profile — a general remedy for impersonation, not an AI-specific penalty. In the platform's own words: “will seek further information before taking actions ranging from temporarily restricting to permanently disabling profiles.”
  Source: Meta Transparency Center — Authentic Identity Representation policy · observed 2026-09-03
• Not yet confirmed: a fixed, quantified penalty schedule for undisclosed synthetic media of real people, such as a strike system or a specific reach-reduction percentage. The only figures located for claims like this trace to marketing and SEO blogs with no Meta-owned citation, so they are not treated as confirmed Meta policy and are left off this record.
  Source: Not confirmed on a Meta-owned page as of this review · observed 2026-09-03
• Not yet confirmed: any Meta-announced policy change tied specifically to the EU AI Act's Article 50 transparency date of August 2, 2026. Meta's AI info labeling infrastructure predates that date, and no dated Meta revision tied to it was located on a Meta-owned page in this review.
  Source: Not confirmed on a Meta-owned page as of this review · observed 2026-09-03

Change log

Dated policy changes we have documented — from the platform's own announcements or named press. Policies change; this log records what we have seen, when.

• 2020 — Meta's original manipulated-media policy covers only fabricated video that makes a person appear to say something they never said — a narrower rule later replaced by the broader labeling approach announced in 2024.
• 2024-02 — Meta's Oversight Board recommends the company overhaul its 2020-era manipulated-media policy.
  Source: About Meta newsroom
• 2024-04 — Meta announces an expanded labeling and manipulated-media approach, widening scope beyond fabricated speech to cover AI-altered depictions of a person doing something they did not do, effective the following month.
  Source: About Meta newsroom
• 2024-05 — The Made with AI label begins rolling out across a wider range of photo, video, and audio content.
  Source: Meta Transparency Center
• 2024-07-01 — Meta renames the label from Made with AI to AI info, after photographers said ordinary edits carrying standard AI indicators were being flagged as if AI-generated.
  Source: TechCrunch, Jul 1, 2024
• 2024-07 — Meta stops removing content solely for violating the old manipulated-video policy, folding enforcement into the label-and-reduced-distribution approach instead.
  Source: About Meta newsroom
• 2024-09-12 — Label placement changes: content only lightly edited with AI tools moves its label into the post's three-dot menu instead of showing it on the post itself.
• 2025-06-25 — Meta's Oversight Board rules that Meta's AI-labeling enforcement is inconsistent, citing a case where copies of the same manipulated election-related audio clip were not consistently auto-labeled.
  Source: Euronews, relaying Meta Oversight Board case ruling
• 2026-08-02 — The EU AI Act's Article 50 transparency obligations become legally live — a disclosure duty that falls on whoever publishes AI-generated or manipulated content, not a Meta-specific policy change. No Meta-announced revision tied to this date was found on a Meta-owned page in this review.
• 2026-09-03 — Record opened — policies as observed on this date.