Uruguayan TV Program Santo y Seña Uses a Deepfake of Political Candidate Yamandú Orsi Without His Consent

Uruguayan TV program Santo y Seña, hosted by Ignacio Álvarez, used AI to create a virtual representation of political candidate Yamandú Orsi, who declined an appearance. Nevertheless, without Orsi's permission, an AI-generated "Orsi" was shown alongside Andrés Ojeda, a rival candidate who appeared in person on the program.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.1 Disinformation, surveillance, and influence at scale

The AI system was used to generate a simulated political interview of a presidential candidate during an election campaign, aiming to influence public opinion and the political process.

Additional risk subdomains

• 3.1 False or misleading information: The creation of a simulated interview carries a high risk of generating and spreading false or misleading information if clips are shared out of context.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The incident was caused by the intentional decision of the television producers to create and broadcast the AI representation of the candidate.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Risk Level 3: Limited Risk. The incident involves the creation of AI-generated content (a deepfake/voice clone), which falls under the transparency obligations of the EU AI Act requiring disclosure that the content is synthetic.

AI system and alleged parties

• AI system: unspecified AI
• AI purpose: Deepfake Video Generation; Voice Generation
• Behaviour type: Tool
• Alleged developer: Unknown deepfake technology developers
• Alleged deployer: Santo y Seña, Canal 4
• Alleged harmed parties: Yamandú Orsi, Uruguayan general public, Uruguayan electorate, Journalism, Electoral integrity, Democracy

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Minor, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Minor
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Minor
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: Yes, the reports describe the creation of an unauthorized AI-generated representation and voice clone of candidate Yamandu Orsi.

Directly caused: The TV show Santo y sena created and broadcasted a simulated virtual avatar and voice clone of Yamandu Orsi.

Inferred additional harm: It is likely that clips of this simulated interview were shared on social media platforms, potentially exposing thousands of viewers to the synthetic content.

Democracy

Reported: Yes, political figures expressed deep concern that the use of AI to simulate candidate interviews damages democratic norms and electoral integrity.

Indirectly caused: The broadcast of the AI-generated candidate alongside a real opponent was described by campaign officials as a 'grave attack on democracy' and a form of 'dirty campaigning' that could undermine the electoral process.

Inferred additional harm: The incident likely contributed to increased distrust in broadcast media and political campaigns among the Uruguayan electorate.

Epistemic

Reported: Yes, the reports highlight the risk of the simulated interview being clipped and spread as genuine statements, eroding shared truth.

Directly caused: The TV show generated a simulated interview where the AI-cloned candidate spoke phrases he did not actually utter in that context.

Indirectly caused: Campaign officials feared that edited clips of the virtual candidate would go viral on social media as real news, spreading misinformation.

Inferred additional harm: Some viewers may have been misled into believing the candidate actually participated or made those specific statements, despite the on-screen disclaimer.

People affected

• Occurrences reported: 1
• People reportedly harmed: 1
• People reportedly exposed: 100000

Potential causes

Management

• Prioritizing Drama over Ethics: Show producers prioritized ratings and controversy over journalistic ethics.
• Inadequate Risk Assessment: Management failed to foresee the political and democratic backlash.

Technology

• AI Voice and Image Synthesis: Allowed the creation of a realistic virtual avatar and voice imitation.
• Technical Glitches in AI Render: Avatar levitated and had rendering errors during the live broadcast.

Data Inputs

• Use of Past Speeches for Training: Producers used historical audio of the candidate to train the model.
• Lack of Consent for Data Use: Candidate voice and image were used without his explicit authorization.

Human Factors

• Candidate Refusal to Attend: Orsi decision not to attend prompted the producers to find an alternative.
• Host Retaliation and Provocation: Host used the AI to mock the candidate for not appearing on the show.
• Public Misinterpretation Risk: Viewers might believe the simulated statements were actually spoken live.

Process and Methods

• Unprecedented Use of AI Simulation: No established protocol for simulating political candidates on live TV.
• Inadequate Disclaimer Layout: Visual disclaimer was used but did not prevent potential clip viralization.

Regulatory Environment

• Lack of AI Campaign Regulations: Uruguay lacked specific legal rules governing AI use in political campaigns.
• No Precedents for Media AI Use: Absence of ethical standards or legal precedents for media-generated avatars.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide clear, consistent accounts of the TV show's broadcast, the reactions from the candidate's campaign, and the specific context of the Uruguayan election. The facts of how the AI was used (voice cloning and avatar generation) and the public controversy are well-documented across multiple sources.
• Ambiguities identified: The specific AI software or platform used by the TV producers to generate the voice and avatar is not named.
• Alternative interpretations: The incident could be viewed purely as a journalistic stunt or satire rather than a malicious disinformation campaign, especially given the on-screen disclaimer.

A Uruguayan television program used AI to generate a virtual avatar and voice clone of presidential candidate Yamandu Orsi after he declined an interview. While the incident sparked controversy regarding electoral integrity and democratic norms, its national security impact remains minor as it was a domestic media stunt with on-screen disclaimers rather than a hostile foreign operation.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: Uruguay
• Alleged perpetrator: Santo y sena producers

Threat characteristics

• Imminence: Near-term. The incident occurred during an active election campaign, presenting near-term concerns for electoral integrity and potential social media spread of misinformation.
• Autonomy: Human-controlled. The AI system was used strictly as a generation tool under the direct control and intent of the television producers.
• Novelty: Evolved capability. While political deepfakes are established, using a real-time virtual avatar to replace a non-consenting candidate on a live television broadcast represents an evolved capability.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or human safety threats were involved in this domestic television broadcast.
• Information security: Minor. The incident involved an AI-generated clone of a presidential candidate. While raising concerns about electoral misinformation, it was a domestic media stunt with an on-screen disclaimer rather than a hostile foreign influence operation.
• Sovereignty: Minor. The simulation of a presidential candidate during an election campaign touches on democratic norms, but did not disrupt core government functions or electoral administration.
• Economic security: Negligible. No strategic technology theft, financial system manipulation, or economic security threats occurred.
• Societal stability: Minor. The deepfake generated public controversy and concerns over 'dirty campaigning,' but did not trigger civil unrest, mass surveillance, or systemic human rights violations.