Meta’s watchdog is investigating apparently AI-generated videos of British politicians misrepresenting their views on issues such as immigration. The commission said the video in question was part of an album posted on Facebook in November that appeared to be a deepfake of a Labor MP representing an area in Scotland making sarcastic and offensive comments about refugees and sexual assault.
In addition to the deepfake video, the post also includes a video of pro-Palestinian protesters that the surveillance committee suspects was AI-generated, as well as a possibly real still image of several women, including the politician from the first video, holding anti-far-right signs. The posts also mention the women’s names and the captions are said to include baseless accusations of tax evasion against Labor politicians.
The board said engagement with the post was relatively minimal and did not include any AI labels or disclosures. It added that two users claimed the content violated Meta’s bullying and harassment policy, but one of them appealed to the board because the company’s systems were unable to escalate the complaint for human review and the content remained on Facebook.
The case summary further explains that the MP in question was speaking out against racist misinformation that was being used as a weapon to stir up anti-immigration sentiment during protests over the accommodation of asylum seekers in UK hotels. They have previously reported threats and intimidation tactics used against them for speaking out about the issue, including AI-generated online slurs.
The oversight committee claims that Mr. Mehta argued that the post was not flagged for deletion or recognized as violating the rules because politicians are public figures of adult age and therefore not automatically protected from “undesirable manipulated images.” Individuals can self-report harmful content and request its removal.
Meta also disputed claims that the content made general statements about refugees that violate its hateful conduct policy, saying it did not meet the standard for misinformation because it had not been reported by members of its “Trusted Partners” program. Additionally, the city councilor depicted was not a candidate in Scotland’s recent local elections, so it could not be considered election interference. Mehta cited the video’s low engagement and “satire” nature as reasons why there was no need to apply the AI label to the video.
The oversight board is currently soliciting public comments related to the case and may make policy recommendations against Meta. These are non-binding, but Meta must respond within 60 days. As a reminder, the Oversight Board is an independent external body, even though it is ultimately financially dependent on Meta.
Meta’s apparent indifference to deepfakes of Scottish politicians is even more surprising given the uproar it got into last year over an AI-generated video of (successful) Irish presidential candidate Catherine Connolly. The video, in which the current Irish president appears to announce that he is withdrawing from the campaign, was shared nearly 30,000 times on Facebook before being deleted.
Connolly called the video a “disgraceful attempt to mislead and undermine voters.” [Ireland’s] Democracy,” Mehta immediately deleted the account where he was hosted. Irish independent. In this case, the video is It was This isn’t the first time Meta has struggled to deal with the emergence of politically motivated deepfakes that have been ruled to violate its community standards.
