Independent research study — not client work

Verified

RS-01Elections and democracy • checked 4 September 2026

AI and election deepfakes: where disclosure is actually enforceable

A short study of the legal instruments themselves, not of press coverage about them. It records what two enforceable obligations actually say — the EU transparency article and one US state statute, both read in full — and marks the wider US state-level picture as only partially verified because it relied on a legislature-association tracker rather than fifty individual statutes.

Section 01

The question and the answer

Question

Which jurisdictions have introduced an enforceable duty to disclose AI-generated political or synthetic media, and what exactly does that duty require?

Answer as published

In the European Union the duty is enforceable and general: Article 50(4) of Regulation (EU) 2024/1689 obliges deployers of AI systems producing deepfake image, audio or video content to disclose that it was artificially generated or manipulated, subject to law-enforcement and artistic-work exceptions. In the United States there is no equivalent federal disclosure statute; the requirements are state-level and uneven. One of those state requirements — Florida Statutes § 106.145 (2024) — was read in full here: it mandates a specified disclaimer on certain AI-generated political content and makes non-compliance a first-degree misdemeanour.

Section 02

Scope

What this study covers, stated before the findings so the limits are not read as an afterthought.
  • Three instruments only: one EU regulation, one US state statute read directly (Florida), and one tracker of the wider US state picture. Nothing here should be read as a global survey.
  • Legal instruments in force at the date of checking. Draft bills, voluntary codes and platform policies are excluded.
  • No claim is made about how often the obligations have been enforced, because enforcement records were not examined.

Section 03

Method as actually run

  1. 01

    Open the consolidated Official Journal text of Regulation (EU) 2024/1689 and read the transparency article in full rather than relying on an explainer site.

  2. 02

    Extract the operative sentence and its exceptions verbatim before writing any summary of it.

  3. 03

    For the United States, look for a single federal disclosure statute; where none is found, say so rather than substituting state examples as if they were national.

  4. 04

    Mark any claim whose underlying page could not be opened directly as partially verified, and say why.

Section 04

Findings, each mapped to a source

F1Verified

EU law imposes a disclosure duty on deployers of deepfake content, effective as part of the AI Act.

Article 50(4) requires deployers of an AI system that generates or manipulates image, audio or video content constituting a deep fake to disclose that the content has been artificially generated or manipulated.

F2Verified

That duty is not absolute — it carries two carve-outs written into the same paragraph.

The obligation does not apply where the use is authorised by law to detect, prevent, investigate or prosecute criminal offence. Where the content forms part of an evidently artistic, creative, satirical, fictional or analogous work, the obligation is limited to disclosing the existence of the generated content in a way that does not hamper display or enjoyment of the work.

F3Verified

The regulation was adopted on 13 June 2024 and published in the Official Journal dated 12 July 2024.

Those two dates appear on the face of the published instrument, which is cited as Regulation (EU) 2024/1689.

F4Partially verified

The United States has no single federal deepfake disclosure statute of comparable scope; the record is state by state.

The National Conference of State Legislatures records at least fifty deepfake-related bills enacted across the states in the 2024 session. The state count itself was not independently re-tallied here, so this finding is marked partially verified; one of the state laws it covers is verified directly in finding F5.

F5Verified

Florida Statutes § 106.145 (2024) makes the AI disclaimer on certain political content a criminal requirement, not just a civil one.

The statute requires a political advertisement, electioneering communication or similar advert that contains generative-AI content which appears to depict a real person performing an action that did not actually occur — and which was created with intent to injure a candidate or to deceive regarding a ballot issue — to state prominently: "Created in whole or in part with the use of generative artificial intelligence (AI)." It prescribes format rules per medium (at least 12-point bold type in print, at least 4 percent of vertical picture height on video, an audible spoken disclaimer for audio), and provides that a person identified as paying for, sponsoring or approving such an advert who omits the disclaimer commits a misdemeanour of the first degree. Any person may file a complaint with the Florida Elections Commission, which must provide an expedited hearing.

Section 05

Sources opened

Every source below was reached at the date shown. Where a page could not be fetched directly, that is stated.
S1Verified

Regulation (EU) 2024/1689 (Artificial Intelligence Act), Official Journal text

European Union — EUR-LexAdopted 13 June 2024; OJ 12 July 2024

Full HTML text retrieved and Article 50(4) read in place; the wording quoted on this page is taken from that text.

Open source
S2Partially verified

Deceptive Audio or Visual Media ('Deepfakes') 2024 Legislation

National Conference of State Legislatures2024 legislative session summary

The page opens normally in a browser but returned 403 to a direct automated request, so its text was read through a search index rather than fetched from the site. Only one of the state statutes it tracks (Florida) was opened individually.

Open source
S3Verified

Florida Statutes § 106.145 — Use of artificial intelligence (2024)

Florida Senate — Online Sunshine statutes2024 statutes (chapter 106, campaign financing)

Full statute text fetched from the Florida Senate site and read directly; the disclaimer wording, format rules and first-degree misdemeanour penalty in finding F5 are quoted from that text.

Open source

Section 06

Search log

QueryEngineLoggedOutcome
EU AI Act Article 50 deepfake disclosure obligation official textWeb search4 September 2026Returned explainer sites first; the Official Journal text was reached and used instead.
site:eur-lex.europa.eu regulation 2024/1689 consolidated textDirect fetch4 September 2026HTML retrieved; Article 50(4) located and read in full.
United States federal deepfake election disclosure lawWeb search4 September 2026No federal disclosure statute of equivalent scope surfaced; state-level tracking material surfaced instead, which set the boundary of finding F4.
Florida 2024 artificial intelligence political advertising disclaimer statuteWeb search, then direct fetch (flsenate.gov)4 September 2026A first guess at the official statute URL failed (HTTP 404, twice, on an unrelated site path); a fresh search located Florida Statutes § 106.145 on the Florida Senate site, which was fetched and read in full for finding F5.

Section 07

What this study does not establish

Stated plainly, because a short study answering a narrow question is only useful if its boundary is visible.
  • How many disclosure duties have actually been enforced, and with what penalties, in any jurisdiction.
  • Whether any platform removal requirement is separately enforceable in either jurisdiction.
  • The position in any jurisdiction outside the EU and the US, including Ghana.
  • The text of any US state statute other than Florida's § 106.145, and whether any of those laws survived constitutional challenge — none of those cases were read.