Meta, the parent company of Facebook and Instagram, has developed a tool to identify and label images generated by artificial intelligence (AI), with the intent of mitigating the spread of AI-created false content. The company acknowledges that the technology is still under development but hopes it will encourage the industry to address the challenge of AI-generated fakes.
Meta's AI image detection system aims to create momentum and incentives for the industry to tackle the issue proactively. However, experts have expressed concerns that such detectors may be easily evaded with minimal image processing and can also generate false positives.
While Meta's tool is limited to images, the company is urging users to manually label their audio and video posts, with potential consequences for failing to do so. This approach has been criticized, considering that the primary concern regarding AI-generated fakes lies with audio and video content.
Meta's current policy on manipulated media has come under scrutiny after a video depicting President Joe Biden behaving inappropriately with his granddaughter was deemed not to violate the guidelines because it did not use AI manipulation. This incident has highlighted the inadequacy of Meta's existing policy to address the increasing prevalence of synthetic and hybrid content.
In response to these concerns, Meta has implemented a policy since January that requires political advertisements to disclose the use of digitally altered images or videos.