Facebook and Instagram owner Meta has said that it will implement new technology to detect and label images generated by other companies’ AI tools.
Set to be deployed across social media networks Facebook, Instagram and Threads, Meta has said that it hopes the new technology will help build ‘momentum’ in the fight against fake AI content.
Meta’s tools already label AI images created by its own systems, and in a blog entry written by senior executive and former British deputy prime minister Nick Clegg, the firm has expressed its desire expand its labelling and detection of fake AI across 2024.
Speaking to the Reuters news agency, Clegg did however admit that the technology was “not yet fully mature”, but that it was underwritten by a desire for the company to “create a sense of momentum and incentive for the rest of the industry to follow”.
Subscribe to Marketing Beat for FREE
Sign up here to get the latest marketing news sent straight to your inbox each morning
Director of the Reliable AI Lab at the University of Maryland, Professor Soheil Feizi struck a downbeat tone when speaking to the BBC, pointing out that such technology would be easily evadable: “They may be able to train their detector to be able to flag some images specifically generated by some specific models.
“But those detectors can be easily evaded by some lightweight processing on top of the images, and they also can have a high rate of false positives. So I don’t think that it’s possible for a broad range of applications.”
The inability of the tools to detect fake audio and video content will equally lead many to question the efficacy of this new technology and whether it will be even remotely adequate in combating the rapid proliferation of AI-generated content.
In his interview with Reuters, Clegg also revealed that the tools would not be able to detect texts generated by tools such as ChatGPT, openly admitting that “that ship has sailed”.
