Are AI-generated evidence and deepfakes covered by rules?
Evidence rules or guidance on the authentication and admissibility of AI-generated material, including deepfakes.
As of 2026-09-04, 1 of 130 countries and entities have a binding rule for ai-generated evidence, 4 have official guidance, 1 have a proposal, 3 are set by case law, 112 were checked with nothing found and 9 remain unclear. Binding countries: Australia.
Binding ruleOfficial guidanceProposed / consultationSet by case lawNothing foundNot yet determined
Tap or hover a country. Faint land has no record yet. Miller projection; boundaries as published by Natural Earth (India point of view) — see the disclaimer.
Affidavits, witness statements and character references must be the maker's own words; Gen AI must not generate their content (NSW SC Gen 23, NCAT PD 7, PIC PD 13). Queensland SC PD 14 of 2024 requires expert reports in criminal proceedings to name the Gen AI program and annex complete prompt and output records.
Arbitration soft law only: SVAMC Guideline 5 asks parties to identify AI-generated material and its limits; Ciarb art. 6 covers admissibility, bias and source-data linkage; VIAC para 6 leaves admissibility and weight to the tribunal; the SCC Guide urges marking of AI-generated content. No international-court evidence rule found.
No evidence rule. The judicial guidelines warn that generative AI can fabricate convincing images, audio and other media which parties could present as evidence, and list red flags for AI-produced submissions, but the Evidence Act 2006 is unamended for synthetic media.
The court guide prohibits using generative AI to fabricate or alter evidence, and the MinLaw guide treats court documents as a high-risk use requiring human review. The Evidence Act 1893 has no AI-specific or deepfake authentication provision.
DIFC PGN 2/2023 requires witness statements to remain in the intended witness's own words under RDC 29.24, addressing AI-drafted testimony. No wider rule on AI-generated or deepfake evidence.
Proposed Federal Rule of Evidence 707 would apply Rule 702 reliability standards to machine-generated evidence offered without an expert. Comment closed 16 Feb 2026; the Standing Committee declined to recommend adoption on 3–4 June 2026 and returned it for further study with deepfake issues.
No AI-specific evidence rule; courts have sanctioned AI-fabricated evidence in Dawu County in November 2025 and Taizhou in December 2025. The Cyberspace Administration labelling measures in force from 1 September 2025 require explicit and metadata labels on AI-generated content, which aids authentication.
In the Wrocław case the court held that printouts of a party's ChatGPT queries submitted as exhibits are not evidence within the meaning of the Code of Civil Procedure. No general rule on AI-generated or deepfake evidence.
Azerbaijan criminalized some unconsented AI deepfakes and requires public labels for AI-generated photo, video and audio materials, but no court evidence admissibility rule was located.
No rule, guidance or reported case found as of 4 September 2026. No Bahraini evidence rule or guidance on authenticating AI-generated or deepfake material was located.
No rule or guidance on authentication or admissibility of AI-generated or deepfake evidence in civil or criminal proceedings found as of 4 September 2026.
No rule, guidance or reported case specifically on authentication or admissibility of AI-generated or deepfake evidence found as of 4 September 2026. Ontario's O. Reg. 384/24 addresses the authenticity of cited authorities, not evidence generally.
No evidence rule or guidance on AI-generated or deepfake material found as of 4 September 2026; Acuerdo PCSJA24-12243 bars judges from using AI to assess evidence, which is a different question.
No rule; the EU-assisted legislative review of June 2026 targets deepfakes as a criminal-law matter. No rule, guidance or reported case found as of 4 September 2026.
No rule; the protocol bans AI translation or analysis of Kichwa and Shuar testimony, which is a procedural safeguard rather than an admissibility rule. No rule, guidance or reported case found as of 4 September 2026.
No rule, guidance or reported case found as of 4 September 2026. The March 2026 governance framework guide places deepfakes in Tier 3 with transparency and watermarking expectations, but that is non-binding and is not an evidence rule.
Evidence rules are national or Member-State procedural law; the EU has no rule on authenticating AI-generated or deepfake evidence. AI Act Art. 50 marking/watermarking duties may assist authentication indirectly but are not evidence rules.
No German evidence rule or guidance on AI-generated or deepfake material found as of 4 September 2026; the free evaluation of evidence under § 286 ZPO applies.
No AI-specific or deepfake-specific evidence rule or guidance found as of 4 September 2026. The Bar Council of India circular of 17 July 2026 prohibits advocates from creating or circulating deepfake or synthetic courtroom content, but that is a conduct rule rather than an evidence rule.
Electronic evidence is governed by the Electronic Information and Transactions Law (UU 11/2008, amended by UU 19/2016 and UU 1/2024). There is no AI- or deepfake-specific authentication rule; the question is discussed only in academic literature.
No Irish rule on authenticating AI-generated or deepfake evidence. PD HC 142 regulates the preparation of witness statements and expert reports but does not address admissibility of AI-generated material as evidence.
No evidence rule or judicial guidance on AI-generated or deepfake material found as of 4 September 2026. A Haifa District Court refusal of a partly AI-produced summary of medical records is described in secondary commentary but was not verifiable against a docket.
The AI Law requires marking of synthetic AI outputs, but no rule on authentication or admissibility of AI-generated evidence in Kazakh courts was located.
No AI-specific or deepfake evidence rule. Section 90A of the Evidence Act 1950 on computer-produced documents is unamended for AI. Caution: the Attorney General's Chambers statute portal is JavaScript-driven and the section text could not be opened, so the statutory position is inferred rather than read.
No rule, guidance or reported case found as of 4 September 2026. The CNDP has invoked article 447-2 of the Penal Code to require labelling of AI-generated political content, but no evidence or authentication rule for AI material in litigation was located.
No evidence rule or guidance on AI-generated or deepfake material. The Rechtspraak AI strategy expressly flags AI-generated process and evidence documents and equality of arms as questions still to be worked out.
No AI-specific evidence rule. Computer-generated evidence is handled under s.84 Evidence Act 2011; commentary argues it is inadequate for generative-AI material, but no amendment or practice direction has been made.
Norway applies free evaluation of evidence [fri bevisbedømmelse]; no rule or guidance on AI-generated or deepfake evidence found as of 4 September 2026.
The 2026 AI law and reporting address synthetic-content labelling and AI-model governance, not authentication or admissibility of AI-generated evidence in court.
No evidence rule on AI-generated or deepfake material. SDAIA's Deepfakes Guidelines address detection and developer obligations but are not rules of evidence and are not addressed to courts.
No evidence rule or guidance on AI-generated or deepfake material found as of 4 September 2026; authenticity is handled under the Electronic Communications and Transactions Act and common-law principles.
No AI-specific evidence rule. The judges' research-group guideline of February 2025 says existing civil and criminal procedure law governs AI-generated evidence and flags deepfake risks. AI Basic Act article 31 requires labelling of generative outputs indistinguishable from reality.
Sweden applies free evaluation of evidence [fri bevisprövning]; no rule or guidance on authenticating AI-generated or deepfake evidence found as of 4 September 2026.
No AI-specific evidence rule; Judicial Weekly in May 2026 recommends adopting the approach of the draft United States Federal Rule of Evidence 707. AI Basic Act article 5 requires warnings on high-risk AI products.
No evidence rule or guidance on AI-generated or deepfake material found as of 4 September 2026. Bills penalising unlabelled AI-generated content have been tabled but none is in force.
Finland applies free evaluation of evidence [vapaa todistusharkinta, OK ch. 17]. No rule or guidance on authenticating AI-generated or deepfake evidence was located.
Not established. No Hong Kong evidence rule or judicial guidance on authenticating AI-generated or deepfake material was located before the session search budget was exhausted.
No AI or deepfake-specific evidentiary rule; the 2001 Rules on Electronic Evidence (A.M. No. 01-7-01-SC) still govern. The Framework only flags judicial training on litigants' use of AI tools or outputs as evidence.
Not substantiated; no AI or deepfake evidence rule located.
Cite this page
SafeLegalAI Legal AI Regulation Tracker, "AI-generated evidence" by country, https://safelegalai.com/regulation/category/evidence-admissibility (accessed 2026-09-04). Data: CC BY 4.0.
SafeLegalAI is a research publication by Cognesio LLP, not a law firm. Nothing here is legal advice, and no lawyer–client relationship arises from reading it. Rules change; always check the official document linked on each record and take advice on your own situation. Researched and drafted with AI assistance; verified against primary sources and edited by Cognesio LLP. The linked official documents are the record — our summaries are not the law and are not legal advice.
Australia
International bodies
New Zealand
Singapore
United Arab Emirates
United States
China
Poland
United Kingdom
Afghanistan
Albania
Algeria
Argentina
Armenia
Austria
Azerbaijan
Bahrain
Bangladesh
Belarus
Belgium
Bolivia
Bosnia and Herzegovina
Botswana
Brazil
Brunei
Bulgaria
Cambodia
Cameroon
Canada
Chile
Colombia
Costa Rica
Côte d'Ivoire
Croatia
Cuba
Cyprus
Denmark
Dominican Republic
Ecuador
Egypt
El Salvador
Estonia
Ethiopia
European Union
France
Georgia
Germany
Ghana
Greece
Guatemala
Honduras
Iceland
India
Indonesia
Iran
Iraq
Ireland
Israel
Italy
Jamaica
Japan
Jordan
Kazakhstan
Kenya
Kuwait
Kyrgyzstan
Laos
Latvia
Lebanon
Lithuania
Luxembourg
Macao
Malaysia
Malta
Mauritius
Mexico
Moldova
Mongolia
Montenegro
Morocco
Mozambique
Myanmar
Namibia
Nepal
Netherlands
Nicaragua
Nigeria
North Macedonia
Norway
Oman
Pakistan
Panama
Paraguay
Peru
Portugal
Puerto Rico
Qatar
Russia
Rwanda
Saudi Arabia
Senegal
Serbia
Slovenia
South Africa
South Korea
Spain
Sri Lanka
Sweden
Switzerland
Taiwan
Tanzania
Trinidad and Tobago
Tunisia
Türkiye
Uganda
Ukraine
Uruguay
Uzbekistan
Venezuela
Zambia
Zimbabwe
Czechia
Finland
Hong Kong SAR
Hungary
Philippines
Romania
Slovakia
Thailand
Vietnam