Alleged Deepfake Audio Depicts Philippines President Ferdinand Marcos Jr. Ordering Military Action

A purported deepfake audio clip portraying Philippine President Ferdinand Marcos Jr. ordering an attack on China spread online in July 2024, fueling tensions in the West Philippine Sea. The Presidential Communications Office reportedly debunked it as fake, attributing it to foreign actors.

Risk classification

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

The deepfake was deployed as part of a coordinated effort to influence public opinion and national foreign policy on a geopolitical scale.

Additional risk subdomains

• 3.1 False or misleading information: The AI system was used to generate false and misleading information regarding a presidential military directive.

Causal factors

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

The risk was caused by malicious human actors who intentionally created and distributed the deepfake audio to manipulate public opinion.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The incident involves an 'audio deepfake' which falls under Risk Level 3 as it is an AI-generated deepfake requiring transparency and labeling obligations under the EU AI Act.

AI system and alleged parties

• AI system: unspecified
• AI purpose: Voice Generation; Social Media Content Generation
• Behaviour type: Tool
• Alleged developer: Unknown voice cloning technology developers, Unknown deepfake technology developers
• Alleged deployer: Unknown actors targeting Ferdinand Marcos Jr., Unknown actors
• Alleged harmed parties: Philippines-China relations, Philippines, Government of the Philippines, General public of the Philippines, General public of China, General public, Ferdinand Marcos Jr.

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 Minor, indirect Minor
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Minor, indirect Negligible
• Epistemic: direct Minor, indirect Minor
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: Yes, the reports explicitly describe the creation and spread of a malicious deepfake audio clip.

Directly caused: A deepfake audio clip mimicking President Marcos Jr. ordering military action against China was created and uploaded to YouTube.

Inferred additional harm: The deepfake likely spawned further shares, re-uploads, and commentary across other social media platforms, exposing a larger audience to the malicious content.

Democracy

Reported: Yes, the reports describe attempts to influence politics and manipulate public perception.

Directly caused: The deepfake attempted to fabricate a presidential military directive, undermining the official communications of the head of state.

Indirectly caused: The incident forced government agencies like the PCO to launch campaigns and coordinate responses to protect information integrity.

Inferred additional harm: Continued use of such deepfakes could erode public trust in government announcements and democratic institutions, especially ahead of the 2025 elections.

Psychological

Reported: Yes, the reports explicitly describe panic caused by the deepfake video.

Directly caused: The deepfake audio caused panic among the public and serious concern among government officials.

Inferred additional harm: Widespread anxiety and distress among the Philippine public regarding potential military conflict with China is likely, though unquantified.

Epistemic

Reported: Yes, the reports explicitly describe the spread of fake news, misinformation, and disinformation.

Directly caused: The deepfake falsely portrayed the Philippine leader as calling for the use of force to retaliate against China, distorting reality.

Indirectly caused: The faked audio led to polarized discussions online, with many viewers accepting the misinformation as fact and expressing strong anti-China sentiment.

Inferred additional harm: The proliferation of such high-quality deepfakes contributes to a broader erosion of shared truth and consensus reality, making it harder for citizens to distinguish real directives from fakes.

People affected

• Occurrences reported: 1
• People reportedly harmed: 1
• People reportedly exposed: 10000

Potential causes

Management

• Platform Engagement Focus: Tech companies prioritized engagement metrics over content safety.
• Inadequate Platform Safeguards: Platforms failed to proactively block high-risk political deepfakes.
• Slow Incident Response: Lack of immediate platform action required government intervention to take down.

Technology

• Generative AI Voice Mimicry: AI advances allowed highly realistic replication of the President's voice.
• AI-Assisted Persona Creation: AI-generated profile pictures and avatars helped spread propaganda.
• Algorithmic Recommendation: YouTube algorithms amplified sensational deepfake content for engagement.

Data Inputs

• Repurposed Media Footage: Old military exercise footage was fed into systems to mislead viewers.
• Unlabeled Synthetic Audio: Deepfake audio was uploaded without any digital watermarks or labels.

Human Factors

• Low Media Literacy: Over half of Filipinos struggle to identify fake news online.
• Cognitive Vulnerability: High geopolitical tensions made viewers susceptible to fear-mongering.
• Viewer Confirmation Bias: Polarized viewers accepted and shared the fake news as truth.

Process and Methods

• Reduced Platform Moderation: Social media platforms scaled back fact-checking and moderation teams.
• Delayed Detection Systems: Lack of real-time AI detection tools allowed the video to go viral.
• Reliance on Manual Flagging: The government had to manually identify and report the deepfake video.

Regulatory Environment

• Lack of AI-Specific Laws: The Philippines lacks enforceable regulations targeting deepfake creation.
• Ineffective Platform Policies: Platforms lack legally binding rules to enforce synthetic content labeling.
• Weak Cross-Border Enforcement: Difficulty in prosecuting foreign actors or individual sources of deepfakes.

Information quality

• Classification confidence: High
• Reason for confidence: Multiple independent news reports consistently describe the deepfake audio, the platform it was hosted on, the specific faked statement, and the official government denial. There is high consensus on the facts of the incident.
• Ambiguities identified: The exact identity of the 'foreign actor' or individual creator remains unconfirmed, with conflicting statements on whether it was a state-sponsored entity or an individual.

A malicious actor deployed a deepfake audio clip of Philippine President Marcos Jr. ordering military action against China during a period of high maritime tension. While quickly debunked, the incident demonstrates a substantial threat to information security and sovereignty by using generative AI to fabricate national security directives and manipulate public opinion.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: Philippines
• Other affected: China
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The immediate crisis was resolved by rapid government debunking, but it represents an ongoing strategic concern for future disinformation campaigns.
• Autonomy: Human-controlled. The AI voice cloning system was used strictly as a tool by human operators who created and distributed the content.
• Novelty: Evolved capability. While deepfakes are an established threat, impersonating a head of state to order military action during an active geopolitical dispute represents a significant escalation.

Impact by dimension

• Physical security: Negligible. No direct physical attacks, kinetic actions, or critical infrastructure disruptions occurred as a result of the deepfake audio.
• Information security: Substantial. Represents a substantial information warfare incident using AI-generated deepfake audio of a head of state to manipulate public opinion and potentially escalate geopolitical tensions.
• Sovereignty: Substantial. Directly targeted the authority of the Head of State by fabricating a military directive, threatening to compromise official government communications and foreign policy.
• Economic security: Negligible. No significant economic or technological security impacts were reported from this incident.
• Societal stability: Minor. Caused public panic and concern regarding potential conflict, though it did not lead to widespread civil unrest or systemic human rights violations.