Portuguese Politicians and Journalists Now Protected Against AI Deepfakes on YouTube

The YouTube platform has rolled out a critical expansion of its AI-detection technology to a pilot group of government officials, journalists, and political candidates—a move that addresses growing concerns about deepfakes and synthetic media ahead of major elections worldwide.

Why This Matters

• Political protection: Candidates running in elections can now request removal of AI-generated content misusing their face or voice, a crucial safeguard ahead of the 2026 campaign season.

• Journalist integrity: News professionals gain a tool to combat identity fraud that threatens their credibility and can fuel disinformation campaigns.

• Global rollout planned: While starting in Brazil, France, the United Kingdom, and the United States, YouTube intends to make the service available worldwide by the end of 2026.

• Not automatic censorship: Detection doesn't guarantee removal—parody, satire, and political commentary remain protected under the platform's editorial guidelines.

How the Technology Works

The tool, known as Likeness ID, functions as a facial and vocal recognition system that scans newly uploaded videos for visual and audio matches against registered public figures. Originally launched in October 2025 for creators enrolled in the YouTube Partner Program, the service now extends to those whose likenesses are most vulnerable to weaponized disinformation: government authorities, journalists covering sensitive beats, and politicians in active campaigns.

To enroll, eligible users must submit a government-issued photo ID and record a short video of their face for biometric verification. YouTube has pledged that this data will be used exclusively for identity verification and detection purposes—not to train Google's generative AI models, a reassurance aimed at privacy-conscious participants.

The detection process mirrors YouTube's existing Content ID system, which has long protected copyrighted music and video. When Likeness ID flags a match, the registered user receives a notification and can review the suspect content. If the video violates YouTube's privacy or impersonation policies, the user may formally request its removal. However, the company evaluates each case individually, preserving content that qualifies as legitimate political criticism, humor, or cultural commentary.

What This Means for Residents in Portugal

While Portugal is not among the initial pilot countries, YouTube's plan to expand Likeness ID globally by end of 2026 means Portuguese politicians, journalists, and civil servants will likely gain access within the next year. This timing aligns with the European Union's enforcement of its AI Act, which includes requirements for transparency in synthetic media.

For Portuguese voters and citizens, Likeness ID offers a practical tool for addressing deepfakes that cross from satire into potential fraud. As election cycles approach, this capability becomes particularly relevant for protecting public figures and maintaining trust in media and political institutions.

The Global Context: Deepfakes and Election Security

YouTube's prioritization of Likeness ID reflects widespread concerns in the electoral security community about AI-generated disinformation ahead of major elections. Across multiple democracies, election authorities are developing policies to address the risks posed by synthetic media and deepfake technology.

The rollout of Likeness ID comes as platforms and governments worldwide recognize the need for both detection tools and clear regulations around AI-generated content. YouTube's approach—combining technology with content policy enforcement—represents one industry response to these challenges.

Industry Response to Deepfake Threats

YouTube is among several technology companies developing deepfake detection capabilities. Various approaches exist in the industry, ranging from pixel-level analysis techniques to behavioral authentication methods. However, all detection systems face the fundamental challenge that AI-generated content continues to evolve, requiring ongoing refinement and updates.

A limitation common to most detection tools is that they operate within specific platforms. Content shared via WhatsApp, Telegram, or TikTok would require separate detection mechanisms, leaving gaps in comprehensive coverage across the broader digital ecosystem.

Legal Framework: The EU AI Act

Under the EU AI Act, which takes effect in Portugal and across the bloc, systems handling sensitive personal data like biometric information face strict requirements. The law mandates transparency in AI-generated content and prohibits certain high-risk uses of AI technology.

YouTube's Likeness ID tool must operate within this regulatory framework. The company has emphasized its commitment to privacy and data protection, stating that biometric data is used solely for detection purposes. However, the balance between effective detection and maintaining user privacy remains an ongoing consideration for such tools.

Transparency and public understanding of how these systems work will be important as they roll out. Users in Portugal and elsewhere deserve clarity about the capabilities and limitations of deepfake detection technology.

The Road Ahead for Portugal

YouTube has indicated it will work with media organizations and political institutions in each expansion market. For Portugal, this outreach process will likely involve discussions with established media outlets and political parties to identify early adopters and gather feedback.

The broader implication for residents is straightforward: as AI-generated content becomes more prevalent, the tools to detect and respond to it are becoming more important. Whether through platform tools like Likeness ID or through regulatory frameworks like the EU AI Act, the conversation around protecting public figures and media credibility is accelerating.

For Portuguese citizens—whether public figures or ordinary voters—understanding these emerging tools and policies will be increasingly relevant as technology and electoral cycles intersect in 2026 and beyond.