Battling Deepfakes: Urgent Call for Facebook’s Action

Dive into the urgent plea for Facebook to address deepfakes and manipulated videos threatening electoral integrity. With the 2024 presidential election approaching, the battle against AI-driven disinformation intensifies. Join the fight to safeguard democracy and combat deceptive propaganda.

The independent committee that oversees Meta’s content moderation decisions is pleading with the internet giant to revise its policy on modified videos to include phony or distorted films that have the potential to deceive voters and influence elections as the country approaches the 2024 presidential election.

A manipulated video of President Joe Biden that surfaced on Facebook in May of last year served as the test case.

Video clips that have been digitally generated or manipulated using generative artificial intelligence to make it seem as though people have said things they haven’t are currently prohibited by Meta. However, it doesn’t deal with more obscene clips—also known as “cheap fakes”—that are created using simple editing software or that depict someone acting in a way they didn’t.

The Oversight Board urged Meta to take tough measures against all doctored content, regardless of how it was made or altered, while upholding Meta’s decision to keep the Biden video up on Facebook. It was also advised that Meta make electoral tampering an explicit target of their policy.

The board stated that phony audio is “one of the most potent forms of electoral disinformation we’re seeing around the world,” therefore it is very concerning.

A fraudulent robocall in January encouraged voters in New Hampshire to forego the primary by using Joe Biden’s voice. The New Hampshire Attorney General’s Office is looking into whether the robocall was made intentionally in an effort to discourage voting. It had no effect on the outcome of the primary – Biden won in a landslide – but it illustrated how generative AI could be used to influence an election, critics say.

Michael McConnell, the co-chair of the Oversight Board, stated in a statement that the policy “makes little sense as it stands.” It forbids edited films that purport to show someone saying things they never said, but it does not forbid posts that purport to show someone acting in a way they never did. It exempts other phony content and only applies to videos made with AI.

Meta did not state if it will abide by the recommendations made by the Oversight Board. According to a spokeswoman, the business was considering the suggestions and will provide a public statement within 60 days.

There’s no guarantee that Meta will invest enough funds and resources to enforce whatever modifications it makes to its distorted media policy.

According to McConnell, “the amount of deceptive content is growing, and the quality of tools to create it is rapidly increasing.” “Platforms need to adapt to these changes quickly, particularly in the context of the upcoming global elections where some actors are attempting to deceive the public.”

Meta defended its guidelines for electoral integrity.

The business released a statement saying, “We have about 40,000 people working on safety and security globally, and protecting the 2024 elections is one of our top priorities.” “We continue to lead the industry in integrity efforts, and we apply the lessons we’ve learned with every election to help stay ahead of emerging threats.”

‘A tsunami of disinformation’ swept through the first AI election.

Not just in the US are there a lot on the line. With more individuals than ever able to cast a ballot in 2024, there’s a greater potential than ever that artificial intelligence will be used in voting. And that is grounds for worry.

Election experts have been preparing for the malicious use of deepfakes in the 2024 presidential fight due to the rapid advancements in technology and the lack of control from the government and business sector. These days, almost anyone may digitally produce or modify photos and videos in realistic ways to trick voters.

Similar to other tech firms, Meta has committed to reducing the negative effects of generative AI. However, despite the fact that technology is becoming increasingly advanced, potent, and commonplace, there are still very few regulations controlling its usage.

Regarding the altered video, the initial clip featured Biden going with his granddaughter to the polls in October 2022, when she was voting for the first time. As she directed, Biden put a “I voted” sticker next to her neck and gave her a cheek kiss. However, the looped video gave the impression that Biden was stroking her chest several times. Biden was called a “sick pedophile” in the caption.

Because AI was not used to edit the video and it did not depict Biden saying anything he did not say, Meta decided to leave it up, claiming it did not break any of its guidelines. In response to a video that was slowed down to make then-House Speaker Nancy Pelosi appear inebriated, the corporation took a same stance in 2019—even as Democrats became enraged.

A deepfake task team has been established by Biden’s 2024 campaign to address deceptive propaganda and lies produced by AI.

“In 2024, there will be a disinformation tsunami. We are now witnessing it, and it will only worsen,” stated Darrell West, a senior scholar at the Brookings Institution’s Center for Technology Innovation. “Everyone is expecting a close election, and anything that changes the balance of 50,000 votes in three or four states could have a significant impact.”

Facebook and other social media networks’ policies on manipulated content

The lackluster reaction from the social media sites hosting this stuff worries West.

Since the 2020 election, West claimed, Meta and other large digital companies have relaxed their misinformation policy and let go of employees tasked with regulating false material, rather than stepping up safeguards.

Political advertisements that cast doubt on the validity of the 2020 US presidential election are now permitted on Meta. Ads that cast doubt on the validity of the current or prospective elections are prohibited.

Therefore, West said, “their capacity to deal with it is quite limited at a time when fake videos are becoming rampant.”

Social media companies have the ability to remove, demote, or attach warning flags to bogus electoral information.

In order to guarantee that the policy is “proportionate,” the Oversight Board suggested that Meta stop eliminating manipulated media and instead apply a label alerting users to the content’s major alteration and potential misleadingness, even in the absence of any other policy violations.

Furthermore, it dissuaded Meta from removing content that fact-checkers flag as manipulated or fraudulent without notifying users or offering an appeals procedure.

“Political speech must be unwaveringly protected. This occasionally includes assertions that are contested and even untrue but do not appear to be detrimental, according to McConnell.

Critics accuse Facebook of not doing more to safeguard elections.

Professor Hany Farid of UC Berkeley, an expert in deepfakes and misinformation, receives daily questions concerning bogus photos on the internet, ranging from Trump with pedophile Jeffrey Epstein to Biden in the situation room wearing military fatigues. For this type of dangerous content, he claims that warning labels are “cowardly.”

The ordinary individual, he claimed, ignores or doesn’t care about the warning label, even if it gives Facebook cover. These labels are typically not applied until a video has received millions of views. Furthermore, the video can subsequently be taken by anybody and posted anywhere without attribution.

Farid claims that Facebook has been at fault for this problem for the past 15 years, with its algorithms favoring content that elicits strong emotions.

“When Facebook claims to have these policies and it’s obvious that those policies are in place to maximize their profits, it’s hard to take them seriously,” the man stated.

Experts on elections need laws against deepfakes.

Daniel Weiner, director of the Brennan Center’s Elections and Government Program, thinks that any attempts by social media corporations to control content that has been manipulated or created by artificial intelligence should be accompanied with deliberate regulations and laws created by legislators.

While AI-generated portraits of Biden are readily refuted, what about a local candidate running for the school board or city council?

Sen. Richard Blumenthal, a Democrat from Connecticut, opened a Senate Judiciary Committee hearing on the possible risks of deepfakes last year by reading a script created by ChatGPT and playing an AI-generated tape that resembled his voice.

More than anything else, Weiner stated, “the most recent developments in AI technology have reinforced the need to strengthen fundamental guardrails for our political system.” These issues have previously existed. Even if every deepfake vanished tomorrow, they would still be around. Furthermore, the solutions are frequently not AI-specific. They are about the need for a broader strengthening of democracy.”

Conclusion

As the threat of deepfakes looms large over electoral integrity, the call for action against manipulated content grows louder. Facebook’s response to addressing AI-driven disinformation will play a crucial role in safeguarding democracy. It’s time for concerted efforts, backed by stringent regulations, to combat the rising tide of election-related deception fueled by advanced AI technology.