This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
High-ranking Philippine government officials, including Senator Ronald 'Bato' dela Rosa and Mayor Sebastian Duterte, shared AI-generated 'man-on-the-street' interview videos created with Google's Veo 3 to influence public opinion during the impeachment proceedings of Vice President Sara Duterte. The videos, which featured synthetic avatars expressing political views, were widely mistaken for real interviews, garnering millions of views and sparking a national debate on disinformation and the responsibility of public officials. The incident has been criticized by the Presidential Communications Office for eroding public trust and has prompted calls for stricter AI regulation in the Philippines.
Philippine Senator Ronald "Bato" dela Rosa and Davao City Mayor Sebastian "Baste" Duterte reportedly shared an AI-generated video made with Google's Veo 3, depicting fake students opposing Vice President Sara Duterte's impeachment. The clip was reportedly styled as a street interview and went viral with millions of views. Reportedly, dela Rosa faced backlash for spreading disinformation but defended the video's message, allegedly ignoring its fabrication.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to conduct large-scale disinformation campaigns, malicious surveillance, or targeted and sophisticated automated censorship and propaganda, with the aim of manipulating political processes, public opinion, and behavior.
Also classified as
The viral spread of synthetic 'man-on-the-street' interviews shared by public officials erodes public trust in shared facts and distorts democratic discourse.
The incident involves the coordinated and at-scale dissemination of highly realistic AI-generated political disinformation to manipulate public opinion regarding a high-profile impeachment trial.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The risk and subsequent public deception were caused by the intentional decisions of human creators to generate political deepfakes and human politicians to share them post-deployment.
Added categories peak at 5: Catastrophic (Epistemic Harm, Inferred)
High-ranking Philippine officials used Google's Veo 3 to generate and share synthetic 'man-on-the-street' videos to influence public opinion during the impeachment of Vice President Sara Duterte. This incident represents a notable escalation in political information warfare, where state actors utilize advanced AI to manipulate public trust and interfere with constitutional processes.
The AI system, Google Veo 3, was used as a text-to-video generation tool that produced synthetic video and audio based on direct user prompts without autonomous planning or decision-making.
Limited Risk: The incident involves the creation and dissemination of AI-generated deepfakes, which fall under the Limited Risk category of the EU AI Act, requiring transparency obligations such as labeling synthetic content.
The primary function of Google Veo 3 in this context was to generate highly realistic synthetic videos of non-existent people to be shared on social media platforms.
The reports provide consistent, detailed accounts of the creation, dissemination, and political fallout of the AI-generated videos. The AI tool (Google Veo 3) is explicitly identified, and the reactions of the key political figures are well-documented across multiple reputable news sources.
None.
Technology
Rapidly Evolving AI Video Tools
Veo and other tools generate highly realistic videos with synchronized audio.
Subtle and Hard-to-Spot Watermarks
Veo's watermark was barely discernible, allowing deepfakes to mislead viewers.
Grey Area in AI Safety Policies
Safeguards block real-person deepfakes but miss fake man-on-the-street avatars.
Data Inputs
Lack of Clear AI Labeling
No prominent labels on shared video, except a hashtag in the caption.
Digitally Fabricated Audio-Visuals
Synthesized students, street scenes, and voices mimic real-world contexts.
Inconsistent Visual Artifacts
Garbled text, nonsense logos, and distorted elements in the generated video.
Human Factors
Lack of Verification by Officials
Senator shared the video without verifying its authenticity first.
Confirmation Bias of Lawmakers
Officials shared the video because they agreed with its political message.
Low Media and Digital Literacy
Netizens struggle to distinguish AI-generated content from real videos.
Process and Methods
Lack of Fact-Checking Protocols
No systematic review process before public officials repost online content.
Algorithmic Monetization of Views
Platforms incentivize engagement and sharing of sensational deepfake content.
Anonymous Disinformation Campaigns
PR firms and pages use anonymity to run organized political influence campaigns.
Management
Acceptance of Fake Content
Leadership defended using AI videos so long as no profit was made.
Evasion of Public Accountability
Officials dismissed critics as trolls instead of admitting the video was fake.
Lack of Institutional Standards
No clear code of conduct for government officials regarding sharing AI media.
Regulatory Environment
Absence of Comprehensive AI Laws
The Philippines lacks dedicated legislation to regulate AI and disinformation.
Delayed Election AI Guidelines
Commission on Elections issued guidelines only months before campaign periods.
Failed AI Regulation Bills
Several bills to establish an AI bill of rights failed to pass in Congress.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
