A group of 20 technology firms announced on Friday their collaboration to combat deceptive artificial intelligence (AI) content that could influence global elections this year. The proliferation of generative AI, capable of rapidly generating text, images, and video in response to prompts, has raised concerns about its potential misuse in electoral processes, especially with over half the world’s population expected to participate in upcoming elections.
Signatories to the tech accord, revealed at the Munich Security Conference, include companies involved in developing generative AI models like OpenAI, Microsoft, and Adobe. Social media platforms such as Meta Platforms (formerly Facebook), TikTok, and X (formerly Twitter) are also part of the agreement, acknowledging their responsibility in mitigating harmful content on their platforms.
The accord outlines commitments to collaborate on tools for detecting misleading AI-generated content, conducting public awareness campaigns to educate voters, and taking action against deceptive content. Techniques like watermarking or embedding metadata may be utilized to identify AI-generated content or certify its origin, according to the companies involved.
However, the agreement lacks specific timelines for implementation or details on individual company strategies. Meta Platforms’ President of Global Affairs, Nick Clegg, highlighted the significance of a unified approach, emphasizing the need for shared interoperable solutions to combat deceptive AI-generated content effectively.
Generative AI has already been exploited in political contexts, including a recent incident where a robocall featuring fake audio of U.S. President Joe Biden circulated to voters in New Hampshire, urging them to abstain from voting during the state’s presidential primary election.
While text-generation tools like OpenAI’s ChatGPT are popular, the focus of the coalition will be on addressing the harmful impacts of AI-generated photos, videos, and audio due to the emotional resonance and credibility associated with multimedia formats, as explained by Dana Rao, Adobe’s chief trust officer.
