Meta Platforms Inc. is ramping up its efforts to combat misinformation ahead of the critical election year, announcing plans to label more posts created using artificial intelligence tools on its platforms Facebook, Instagram, and Threads. The move comes as part of Meta’s broader initiative to prevent the spread of deceptive content online, particularly during election seasons.
Meta to begin detecting AI-generated posts from Google, OpenAI
Working in collaboration with other tech companies, Meta is establishing common technical standards for identifying AI-generated posts, including implementing invisible watermarking and metadata to images upon creation. This allows the company to develop software systems capable of detecting these invisible markers and labeling AI-created content, even if it originates from a competitor. Nick Clegg, Meta’s president of global affairs, anticipates the ability to detect and label images generated by various AI-focused companies like Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock in the coming months.
With numerous elections scheduled for 2024 across countries such as the US, India, South Africa, and Indonesia, the threat of disinformation looms large. The proliferation of generative AI tools has made it easier to produce convincing fake images, text, and audio, exacerbating the challenge of combating misinformation. While Meta’s system initially focuses on detecting AI-generated images from other companies’ tools, the company is also working on methods to identify AI-generated audio and video content.
As the upcoming elections, including in the US, draw closer, Meta’s efforts to detect AI deepfakes have taken on increased urgency. The company’s commitment to establishing industry standards around watermarking was underscored by Clegg at the World Economic Forum in Davos last month. Despite the potential for AI-generated content to influence elections if left unlabeled or unchecked, Clegg remains optimistic that the collaborative industry efforts will mitigate such risks.
Addressing concerns raised by experts, Meta’s Oversight Board recently critiqued the company’s manipulated media policy, emphasizing the need for improved labeling of AI-generated content instead of outright removal. Clegg concurred with the board’s assessment, highlighting the significance of the watermarking updates as a step in the right direction. Looking ahead, he emphasized the importance of a broader discussion within the industry to address the authenticity and veracity of both synthetic and non-synthetic content on the web.
