Deepfake of Long-Deceased Suharto Circulating in Run-up to February 2024 Indonesian Elections

An AI-generated deepfake of Suharto, the deceased Indonesian dictator, was generated and circulated by the Golkar Party ahead of the February 2024 Indonesian elections. This video, which aimed to influence voter perceptions by invoking Suharto's legacy, sought to manipulate public opinion and misused deceased individuals' likenesses for political gain. The incident is another example of political deepfakes creating convincing misinformation.

Risk classification

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

The incident involves coordinated, large-scale political campaigns using AI-generated deepfakes and synthetic media to influence public opinion and manipulate voters during a national election.

Additional risk subdomains

• 4.3 Fraud, scams, and targeted manipulation: Candidate Anies Baswedan was targeted with a fabricated audio deepfake designed to manipulate public perception of his political standing.
• 3.1 False or misleading information: The campaigns generated and spread highly convincing false audio and video representations of political figures to mislead the electorate.

Causal factors

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

The risk was caused by human political campaigns intentionally deploying AI-generated deepfakes and synthetic media post-deployment to manipulate public opinion and influence voters.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Risk Level 3: Limited Risk. The report describes the use of 'AI-generated content such as deepfakes' and 'interactive AI chatbots' which require transparency obligations under the Act so that users are aware they are interacting with AI.

AI system and alleged parties

• AI system: OpenAI-powered chatbot, unspecified (OpenAI)
• AI purpose: Deepfake Video Generation; Voice Generation
• Behaviour type: Tool
• Alleged developer: Golkar Party
• Alleged deployer: Golkar Party
• Alleged harmed parties: Indonesian electorate, 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 Severe
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Minor, indirect Negligible
• Democracy: direct Minor, indirect Minor
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Severe
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: The report explicitly describes the creation and spread of malicious political deepfakes and propaganda.

Directly caused: Political parties created a deepfake video of deceased dictator Suharto and a fabricated audio clone of candidate Anies Baswedan to manipulate voters.

Indirectly caused: The Suharto deepfake video spread widely across X, TikTok, Facebook, and YouTube, accumulating over 4.7 million views.

Inferred additional harm: Numerous other unquantified deepfakes and AI-generated campaign materials were likely spread across Indonesian social media networks during the election cycle.

Civil rights

Reported: The report explicitly describes actions taken to circumvent campaign regulations and manipulate voters.

Directly caused: The Gerindra Party used AI-generated children in TV commercials to skirt rules banning real children from appearing in political campaigns.

Inferred additional harm: The use of synthetic media to bypass election regulations and manipulate public opinion may have undermined the civil rights of voters to a fair and transparent election process.

Democracy

Reported: The report explicitly describes the erosion of democratic norms through voter manipulation and deceptive campaigning.

Directly caused: Political parties used deepfakes of a deceased dictator and fabricated audio of candidates to influence voter behavior and manipulate the election.

Indirectly caused: Watchdog groups like TAPP and Access Now expressed deep concern that these deepfakes could sway and manipulate millions of voters, undermining democratic integrity.

Inferred additional harm: The widespread, unregulated use of AI in campaigns likely eroded public trust in the electoral system and democratic institutions in Indonesia.

Epistemic

Reported: The report explicitly describes epistemic harm through the generation of highly convincing false audio and video deepfakes.

Directly caused: An AI-generated deepfake of deceased dictator Suharto and a doctored audio conversation of candidate Anies Baswedan were created and circulated.

Indirectly caused: The Suharto deepfake was viewed over 4.7 million times, potentially misleading viewers, and Anies Baswedan had to publicly warn voters to be critical of AI.

Inferred additional harm: The proliferation of synthetic media makes it increasingly difficult for Indonesian voters to distinguish between real and fabricated political communications, eroding the shared information ecosystem.

People affected

• Occurrences reported: 1
• People reportedly harmed: 1
• People reportedly exposed: 4700000

Potential causes

Management

• Prioritizing Outreach Over Ethics: Political parties prioritized voter engagement over ethical AI considerations.
• Lack of Campaign Accountability: Campaign teams deployed AI without internal ethical review or risk assessment.

Technology

• AI Face and Voice Cloning Tools: Advanced AI tools easily cloned Suharto's face and voice to create deepfakes.
• Generative AI Image Software: Software used to create fake images of children to bypass campaign rules.

Data Inputs

• Historical Audio and Video Footage: Abundant archival data of Suharto enabled highly realistic voice cloning.
• Unlabeled AI Generated Content: Lack of metadata or watermarks allowed deepfakes to pass as real footage.

Human Factors

• High Social Media Consumption: Indonesians' high internet usage facilitated rapid viral spread of deepfakes.
• Voter Susceptibility to Manipulation: Voters easily misled by realistic AI representations of political figures.

Process and Methods

• Bypassing Ad Campaign Rules: Parties used AI children to circumvent bans on children in political ads.
• Rapid Social Media Distribution: Algorithms prioritized viral deepfake content, spreading it to millions.

Regulatory Environment

• Lack of Deepfake Regulations: No specific laws or enforcement mechanisms existed to restrict political AI.
• Slow Government Response: Ministry advisories were insufficient to curb the viral spread of deepfakes.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear, detailed accounts of multiple specific AI deployments in the Indonesian election, including the platforms used, view counts, and statements from political parties, candidates, and watchdog groups. There is minimal ambiguity regarding the events, and the classifications are well-supported by the text.
• Ambiguities identified: The exact AI models used to generate the Suharto deepfake and the Anies Baswedan audio clone are not specified, though OpenAI is mentioned in relation to Anies's campaign chatbot.

Multiple Indonesian political parties deployed AI-generated deepfakes and synthetic media, including a deepfake of deceased dictator Suharto and fabricated candidate audio, to influence voters and bypass campaign regulations during the 2024 election. This represents a substantial information warfare and sovereignty concern, highlighting the evolving threat of AI-driven election manipulation.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: Indonesia
• Alleged perpetrator: Indonesian political parties

Threat characteristics

• Imminence: Near-term. The threat developed over the course of the election campaign, requiring monitoring and response within a sub-30 day window.
• Autonomy: Human-controlled. AI tools were fully controlled and directed by human political campaigns to generate specific synthetic media outputs.
• Novelty: Evolved capability. Represents a significant advancement and scaling of existing deepfake and synthetic media capabilities in a major national election.

Impact by dimension

• Physical security: Negligible. No physical security threats, kinetic attacks, or critical infrastructure compromise reported in this incident.
• Information security: Substantial. Widespread deployment of high-fidelity deepfakes targeting political campaigns and voters, generating millions of views and manipulating the information ecosystem.
• Sovereignty: Substantial. AI-generated synthetic media, including deepfakes and simulated children, were intentionally used to influence over 200 million voters and circumvent campaign regulations during a national election.
• Economic security: Negligible. No direct threats to economic stability, financial systems, or strategic technological assets reported.
• Societal stability: Minor. Circumvention of campaign regulations and proliferation of deceptive media threatened the integrity of the election, though it did not lead to widespread civil unrest.