Adam Mosseri, head of Instagram, admitted that the platform does not yet have an effective solution to deal with profiles using deepfakes created by artificial intelligence to replicate content. This issue has been reported to him by creators within Instagram itself, according to reporting by the British newspaper The Independent.
One of the cases reported involved American comedian George Brett Olson, who identified an account using an AI-generated female avatar in place of him during one of his performances. Although the original audio and text were maintained, the image was replaced by this deepfake.
The social media executive expressed great concern about this practice but opted not to completely ban content produced with the technology, indicating that he will discuss possible measures with the platform's team.
Olson reported that this type of material has appeared in other areas, such as profiles that use videos of dancers as a basis to create recordings where AI characters reproduce choreographies.
The profile that imitated the comedian's performance had over 300 thousand followers and directed users to an OnlyFans page. Another account, mentioned by comedian Julia Jasunas, had approximately 250 thousand followers.
Mosseri responded directly to Olson's post on Instagram, stating that Meta is aware of these profiles, but that a broad restriction on AI content would not be feasible. He justified this by saying that 'it will become increasingly difficult to identify, and not all content is bad, although there is a lot of junk.'
Furthermore, Mosseri declared that Instagram plans to direct more traffic to original creators and acknowledged the platform's need to define how to proceed when content is an exact copy of these authors' productions. He cited examples on competing networks, such as Elon Musk's X, where he changed his recommendation algorithm to restrict the reach of copied content and modified the monetization system to favor original creations.
In another comment, Mosseri addressed the tools available for requesting material removal. The platform offers reports for copyright infringement, but the executive observed that this mechanism does not always resolve the situations raised by creators.
According to him, many creators may not have adopted formal procedures to protect their work, and the network needs to find a way to 'help these creators without allowing malicious actors to make more false reports,' something that, according to Mosseri, 'happens all the time.'
Previously, Mosseri had debated the dilemma of handling AI on the platform. He had previously said that content generated by artificial intelligence should not be shown to those who do not show interest in it, but denied that users had control to block it, keeping this function under the responsibility of the algorithm.