Meta, formerly known as Facebook, is intensifying its efforts to combat the spread of misinformation and deepfakes created by artificial intelligence (AI) ahead of upcoming elections worldwide. In a recent announcement, the company revealed its plans to develop tools that can identify AI-generated content not only from its own AI tools but also from other major AI platforms. This article delves into Meta’s expanded strategy against AI-generated misinformation and the challenges it faces in fighting this technological menace.
Until now, Meta’s focus was on identifying AI-generated images created using its own AI tools. However, the company is looking to widen its scope by applying labels to content generated by AI technologies developed by Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock. This expansion aims to encompass a broader range of platforms and languages, ensuring that users can be alerted to the presence of potentially deceptive content.
Meta acknowledges that implementing this change will take time. Nick Clegg, Meta’s president of global affairs, stated that the company plans to begin labeling AI-generated images from external sources “in the coming months” and will continue working on the issue throughout the next year. The extended timeline is necessary for Meta to collaborate with other AI companies to establish common technical standards for identifying AI-generated content reliably. These standards will indicate when a piece of content has been created using AI.
The 2016 presidential election highlighted the crisis Facebook faced due to the proliferation of election-related misinformation. Foreign actors, predominantly from Russia, exploited the platform to spread highly charged and factually inaccurate content. In subsequent years, Facebook found itself continuously targeted by misinformation, particularly during the Covid pandemic. In preparation for the 2024 cycle, Meta aims to demonstrate its readiness to combat bad actors utilizing advanced AI technology.
While some AI-generated content is easily detectable, many instances remain more challenging to identify. Services claiming to detect AI-generated text have exhibited biases against non-native English speakers. Similarly, identifying AI-generated images and videos is not straightforward, although certain signs can indicate their origin. To minimize uncertainty, Meta plans to collaborate primarily with other AI companies employing invisible watermarks and specific metadata in their images. However, the removal of watermarks poses a significant challenge that Meta intends to address.
Meta is actively working on developing classifiers capable of automatically detecting AI-generated content, even if it lacks invisible markers. The company is determined to enhance its ability to identify deceptive content while also exploring ways to impede the removal or alteration of invisible watermarks. In the case of audio and video, monitoring proves even more daunting as there is currently no industry standard for AI companies to include invisible identifiers. As a partial solution, Meta plans to introduce a feature enabling users to voluntarily disclose when they upload AI-generated video or audio. Failure to disclose such content may result in penalties enforced by the company.
In cases where digitally created or altered image, video, or audio content poses a high risk of materially deceiving the public on matters of importance, Meta may apply a more prominent label. This measure serves to alert users to the potential for manipulation and misinformation, allowing them to approach the content with caution. Meta’s ultimate goal is to protect its user base from falling victim to deceptive AI-generated media.
Meta’s expanded effort to combat AI-generated misinformation and deepfakes demonstrates the company’s commitment to safeguarding its platforms and users from the threats posed by advanced AI technology. By partnering with other major AI companies and establishing common technical standards, Meta aims to strengthen the detection and labeling of deceptive content. While challenges persist, Meta’s dedication to this vital endeavor signals a proactive approach to the battle against AI-generated misinformation in the digital age.