The United Nations’ International Telecommunication Union (ITU) has urged governments and companies to take stronger action against AI-generated deepfake, warning that they pose rising risks to elections, financial systems, and public trust. The call came in a new report released at the AI for Good Summit in Geneva, where experts highlighted the urgent need for digital content verification standards.
The ITU report emphasised that AI-powered images, videos, and audio clips impersonating real individuals are already being weaponised to mislead and manipulate. It called on social media platforms and digital distributors to adopt provenance tools, including watermarking and verification tags, to help users differentiate between authentic and synthetic content.
Experts warn of global vacuum in deepfake governance
Bilel Jamoussi, Chief of the ITU’s Standardization Bureau, stated that public trust in social media is declining rapidly due to the rise of manipulated content. Meanwhile, Adobe’s Leonard Rosenthol noted that verifying the source and timestamp of digital content is crucial for user trust: “People want to know—can I believe this video, this image?”
Also read: Cybercrime Losses in Karnataka Cross ₹2,915 Cr
Cyber governance expert Dr. Farzaneh Badiei added that the absence of a global watchdog or consistent standards makes deepfake threats more dangerous. Fragmented responses across countries could lead to loopholes that attackers exploit at scale, especially during politically sensitive events.
Industry urged to adopt AI safety standards proactively
The ITU is currently developing standards for video watermarking, which would embed metadata like creator identity and creation time. With video now accounting for 80% of all internet traffic, the UN agency is pressing for urgent industry cooperation to adopt such tools across platforms.
Tomaz Levak of Switzerland-based Umanitek emphasised the role of private sector accountability and user education, warning that AI will only become faster and more convincing. Organisations that act early will be better equipped to safeguard users and preserve the integrity of public information ecosystems.
