Latvia - Criminalising Deepfake Use (90.1/2024)
Use of deepfake technology to influence the electoral process
Vēlēšanu procesa ietekmēšana, izmantojot dziļviltojumu tehnoloģiju
Latvia
RAI-LV-NA-ACA9CXX-2024In May 2024 the Saeima adopted amendments to the Criminal Law adding Article 90.1 to criminalise the creation or dissemination of intentionally false, discrediting information using deepfake technologies during the pre-election campaign period or on election day, and to criminalise the use of automated data processing systems to prevent persons from participating in democratic processes. Penalties include imprisonment up to five years and other custodial or non-custodial sanctions. The amendments entered into force on 22 May 2024.
Summary
In 2024 the Saeima of the Republic of Latvia amended the Criminal Law by inserting Article 90.1, titled (in Latvian) "Vēlēšanu procesa ietekmēšana, izmantojot dziļviltojumu tehnoloģiju" (use of deepfake technology to influence the electoral process). The legislative change was adopted in the final (third) reading on 9 May 2024 and entered into force on 22 May 2024. The amendment criminalises the deliberate production or dissemination of false, discrediting audio-visual or audio materials created with deepfake technologies when targeted at political organisations (parties), party coalitions, or candidates to the Saeima, local government councils or the European Parliament, if carried out in the pre-election campaign period (defined in election law as the period from 120 days before the election up to election day) or on election day itself. The law also extends criminal liability to activities that employ automated data processing systems to prevent persons from (a) participating in elections, (b) initiating or participating in referendums, or (c) supporting a European Citizens' Initiative.
The statute responds to growing concerns about the realism and reach of AI-generated content and the risk that such content can undermine voters' ability to make informed choices and the integrity of democratic processes. Maximum sanctions for deepfake-related election offences are set at imprisonment of up to five years (as reflected in the adopted redaction and subsequent committee adjustments), temporary deprivation of liberty, probation supervision, or community service; the law also removes monetary fines as a primary sanction for the specially aggravated use of automated systems to impede democratic participation. The amendments include transitional and clarifying provisions referencing how content created prior to the law's entry into force is to be treated, and they broaden existing Article 90 provisions to encompass automated systems and AI-enabled manipulative practices.
Practically, the law imposes criminal exposure on authors, producers and disseminators (including, where applicable, those who knowingly and intentionally distribute deepfake content) and signals Latvian intent to protect the transparency and fairness of election campaigns. Criminal investigations and prosecutions under the new provisions will be conducted by Latvian investigative authorities and prosecuted by the Prosecutor’s Office; the State Police and, where relevant, the Corruption Prevention and Combating Bureau (KNAB) may be involved for crimes against the state or where political financing or public office is implicated. The law sits alongside other regulatory measures addressing AI use in pre-election campaigning, and with administrative election rules; it is intended to complement administrative restrictions rather than displace them. Relevant official materials and parliamentary documents (Saeima press releases, parliamentary dossier of the bill, and publisher guidance) record the adoption process and provide explanatory notes and transitional rules. Key public sources include the Saeima press release reporting the third reading, the LV Portāls explanatory note and the official official-disseminated session records and promulgation notices.
Full article
Read full text ↗Overview
The Amendments to the Criminal Law (introducing Article 90.1) were adopted by the Saeima on 9 May 2024 and entered into force on 22 May 2024. The amendments criminalise the manufacture and dissemination of intentionally false, discrediting information produced using deepfake ("dziļviltojuma") technologies about political organisations or candidates if the acts occur during the legally defined pre-election campaigning period or on election day. The law also adds criminal liability where such acts are carried out by automated data processing systems aimed at obstructing participation in elections, referendums or European Citizens’ Initiatives. For the official parliamentary announcement see Saeima press release (9 May 2024) and for an accessible summary and entry-into-force information see LV Portāls (22 May 2024).
Definitions
Key statutory definitions are provided or clarified in the drafting and accompanying explanatory notes. "Deepfake"/"dziļviltojums" is understood to include faked or AI/manipulation-generated images, video, or audio that portray a real or realistic-seeming person saying or doing things they did not say or do, or fully synthetic persons created by automated systems. The "pre-election campaign period" is cross-referenced to electoral law and defined as the 120 days prior to election day up to and including election day (with specific rules where elections are re-run or an assembly is dissolved). "Automated data processing system" covers, in the statute's scope, systems using automated algorithms, bots, or other automated means to manage or disseminate content and to influence or obstruct persons’ participation in democratic acts.
Governance and Institutional Framework
Implementation and enforcement fall primarily to criminal justice institutions: investigative functions are exercised by the State Police (Valsts policija) and specialised units where applicable; prosecutions are the responsibility of the Prosecutor’s Office (Ģenerālprokuratūra); and political corruption, financing or high-level interference allegations can engage the Corruption Prevention and Combating Bureau (KNAB). The Saeima remains the lawmaker and source of statutory interpretation through parliamentary committee materials. The law’s explanatory notes and parliamentary dossier (Likumprojekts Nr. 535/Lp14) elaborate institutional responsibilities and interplay with election administration authorities. For the Saeima dossier and legislative documents see the Saeima parliamentary database and session materials archived by the official publication Latvijas Vēstnesis session record.
Key Focus Areas
The statute concentrates on several policy objectives: (1) safeguarding the integrity of electoral decision-making by criminalising materially false, reputation-damaging deepfake content targeted at political actors during the pre-election window; (2) deterring and penalising the use of algorithmic/automated systems to suppress or obstruct voters or civic participation; (3) clarifying sanctions and ensuring proportionality of penalties for offences that threaten democratic institutions; (4) creating evidentiary and procedural expectations for prosecutors and investigators in AI/technology-enabled offences; and (5) providing transitional rules for content created before the law’s entry into force. The statute deliberately targets knowingly false and discrediting conduct rather than all synthetic content. It excludes lawful journalistic, satirical or academic uses where no intentional falsehood aimed at discrediting is present. Related policy materials and analyses discuss the threshold for proof of "knowledge" and the operational challenges in attributing authorship of automated dissemination, see the Saeima explanatory notes and public legal commentary from independent law firms and legal periodicals.
Implementation Framework
Operationalising Article 90.1 requires multi-disciplinary capacity: technical forensic capabilities to detect and preserve evidence of synthetic media; legal training for investigators and prosecutors on digital attribution and expert evidence; clear internal protocols for preservation and chain-of-custody of online material; coordination protocols between police, prosecutor’s office and election administration bodies; and guidance for media platforms and intermediaries on cooperation with law enforcement while respecting freedom of expression. The statute does not create an administrative takedown regime but investigators may seek judicial orders for evidence preservation or removal where criminal processes require. Parliamentary materials and government guidance recommend investment in digital forensics and cross-border cooperation to address content hosted or generated outside Latvia. The Saeima's press release and accompanying committee documentation outline the intended prosecutorial approach and references to existing criminal procedure frameworks.
Monitoring and Evaluation
Monitoring will combine traditional criminal statistics with qualitative review of case law and internal prosecutorial reports. Key performance indicators include number of complaints received, investigations opened, prosecutions commenced, convictions or acquittals, timeliness of digital evidence preservation actions, and cross-border legal assistance executed. Oversight is through regular reporting by prosecutorial and police authorities to the legislature and through public transparency channels. The law’s designers envisaged periodic evaluation to ensure offences are applied proportionally and that enforcement does not chill legitimate expression or journalistic activity. External researchers and civil society are expected to review implementation and help refine prosecutorial guidance.
Penalties, Liability, and Appeals
The statutory penalty range includes imprisonment (up to five years in aggravated or automated-system cases), temporary deprivation of liberty, probation supervision or community service. Monetary fines were limited or removed from specific aggravated provisions to reflect the gravity attributed to attacks on electoral integrity using automated systems. Liability attaches to authors, producers and intentional distributors; where multiple parties are involved the law contemplates joint liability in line with general principles of criminal law. Defendants retain full criminal procedural protections and may appeal convictions through the ordinary appellate courts; issues of evidentiary sufficiency, expert testimony reliability, and jurisdiction for cross-border digital evidence are likely to be focal points in appeals and jurisprudence.
Relationship to Other Instruments
Article 90.1 complements existing electoral law (which defines campaigning periods and administrative restrictions), data protection obligations (where processing of personal data for synthetic content may implicate the Data State Inspectorate), and other criminal statutes (e.g., fraud, defamation, unlawful interference). It was coordinated with proposed amendments in the Criminal Procedure Code to enable digital evidence handling. The amendment is not designed to replace administrative or civil remedies; rather it provides a criminal backstop for serious, intentional misconduct. Practitioners are advised to consult the Saeima legislative dossier (Likumprojekts Nr. 535/Lp14) and explanatory notes for links to related statutory texts and transitional rules.
International Alignment
Latvia’s approach aligns with wider European concern about AI-driven disinformation and electoral interference. The law references precedent jurisdictions and cross-border incidents noted in parliamentary debate (cases in Argentina, Slovakia and Poland were cited in committee materials). Latvia’s measure sits alongside EU-level proposals on AI safety and digital services regulation and may be invoked in international cooperation for cross-border investigations and mutual legal assistance. The Saeima’s explanatory notes reference alignment with democratic protection principles promoted in EU instruments and encourage cooperation with EU law enforcement and digital platform regulation initiatives.
Implementation Timeline
| Event | Date |
|---|---|
| First reading (Likumprojekts Nr. 535/Lp14) | 2024-03-21 |
| Juridical committee conceptual support | 2024-03-13 |
| Adopted in final (third) reading by Saeima | 2024-05-09 |
| Promulgation/entry into force | 2024-05-22 |
| Implementation (technical capacity building & guidance issuance) | 2024 Q3–Q4 (ongoing) |
Sources and References
Requirements for a company
What an organisation has to do under Latvia - Criminalising Deepfake Use (90.1/2024), at a glance. Not legal advice — the table below gives the provision and deadline for each item.
Must do
8- Do not create or spread intentionally false, discrediting deepfakes about political actors.Any person or entity.
- Do not use automated systems to obstruct participation in democratic processes.Any person or entity operating automated data processing systems.
- Audit content creation policies to prevent deepfake misuse during election periods.Organizations creating or disseminating content.
- Advise staff and vendors on deepfake prohibitions and election interference.Organizations creating or disseminating content.
- Disable or reconfigure automated systems that could obstruct democratic participation.Operators of automated data processing systems.
- Implement logging and retention protocols for relevant digital content.Media platforms and intermediaries.
- +2 more in the table below
Must not do
0Nothing in this category.
Should do
1- Conduct Data Protection Impact Assessments for automated systems impacting democratic acts.Operators of automated data processing systems.
Should not do
0Nothing in this category.
Who must do what
The obligations under Latvia - Criminalising Deepfake Use (90.1/2024), most serious first. Not legal advice — verify against the official text before relying on it.
| # | Who | Requirement | By when | Where | Severity |
|---|---|---|---|---|---|
| 1 | Any person or entity. | Do not create or spread intentionally false, discrediting deepfakes about political actors. “criminalise the manufacture and dissemination of intentionally false, discrediting information produced using deepfake... about political organisations or candidates” | May 22, 2024 | Article 90.1 | Critical |
| 2 | Any person or entity operating automated data processing systems. | Do not use automated systems to obstruct participation in democratic processes. “criminal liability where such acts are carried out by automated data processing systems aimed at obstructing participation in elections” | May 22, 2024 | Article 90.1 | Critical |
| 3 | Organizations creating or disseminating content. | Audit content creation policies to prevent deepfake misuse during election periods. “audit content creation policies; advise staff and vendors” | — | — | Important |
| 4 | Organizations creating or disseminating content. | Advise staff and vendors on deepfake prohibitions and election interference. “advise staff and vendors” | — | — | Important |
| 5 | Operators of automated data processing systems. | Disable or reconfigure automated systems that could obstruct democratic participation. “Disable or reconfigure automated outreach systems that could interfere with rights” | — | — | Important |
| 6 | Media platforms and intermediaries. | Implement logging and retention protocols for relevant digital content. “Implement logging and retention protocols” | 2024 Q3–Q4 (ongoing) | — | Important |
| 7 | Media platforms and intermediaries. | Designate officers responsible for digital evidence preservation. “designate preservation officers” | 2024 Q3–Q4 (ongoing) | — | Important |
| 8 | Media platforms and intermediaries. | Develop procedures for cooperating with law enforcement requests. “Develop contact and compliance procedures for law enforcement requests” | 2024 Q3–Q4 (ongoing) | — | Important |
| 9 | Operators of automated data processing systems. | Conduct Data Protection Impact Assessments for automated systems impacting democratic acts. “conduct DPIAs where relevant” | — | — | Recommended |
Related Regulations
Latvia - AI Use in Campaigning (2024)
Latvia94% similar
Germany - Protection Against Deepfakes (222/24)
Germany90% similar
United States - Minnesota - Deepfakes Regulation (HF 4772)
United States89% similar
Latvia - AI Regulation Overview
Latvia89% similar
United States - South Dakota - Deepfakes Regulation (SB 164)
United States89% similar
© Regulations.AI · updated on 13-Jun-2026