Big tech companies signed an agreement Friday to voluntarily adopt “reasonable precautions” to prevent artificial intelligence tools from being used to interfere with democratic elections around the world.
Technology executives from Adobe, Amazon, Google, Meta, Microsoft, OpenAI, and TikTok gathered at the Munich Security Conference to announce a new voluntary framework for how to respond to AI-generated deepfakes that intentionally mislead voters. Thirteen other companies have also signed the agreement, including IBM and Elon Musk’s X.
“Everyone recognizes that no tech company, no government, no civil society organization can address the emergence of this technology and its potential for abuse alone,” said Nick, Meta’s global president.・Mr. Clegg said. He spoke in an interview ahead of Summit, the parent company of Facebook and Instagram.
Although largely symbolic, the agreement targets increasingly realistic images, audio, and video generated by AI, and is intended to help “political candidates, election officials, and other key stakeholders in democratic elections.” “Deceptively disguises, alters, or otherwise provides false information to voters about when, where, or how they can legally vote.” become. ”
The companies have no intention of banning or removing deepfakes. Instead, the agreement outlines methods to be used to detect and label deceptive AI content as it is created or distributed on the platform. The magazine said the two companies will share best practices with each other and will take “quick and appropriate responses” if the content begins to spread.
The vagueness of the pledge and lack of binding requirements may have helped win support from a wide range of companies, but may disappoint pro-democracy activists and watchdog groups seeking stronger guarantees.
“It makes sense that every company has its own content policy,” says Clegg. “This is not an attempt to put a straitjacket on everyone. And in any case, no one in the industry should sweep it under the rug, play whack-a-mole and potentially mislead anyone. We don’t think we can deal with a completely new technological paradigm by figuring out everything that seems to be there.”
The agreement at the German city’s annual security conference comes as more than 50 countries are planning national elections in 2024. Some countries have already done so, including Bangladesh, Taiwan, Pakistan, and most recently Indonesia.
Attempts to use AI to interfere in elections have already begun, such as when an AI robocall mimicking US President Joe Biden’s voice tried to dissuade people from voting in the New Hampshire primary last month.
Days before Slovakia’s general election in November, an AI-generated audio recording impersonated a liberal candidate discussing plans to raise beer prices and rig elections. Fact checkers scrambled to identify them as false, but they had already been widely shared as real on social media.
Politicians and election commissions have also been experimenting with the technology, including using AI chatbots to communicate with voters and adding AI-generated images to ads.
Ahead of Indonesia’s elections, the leader of a political party has shared a video replicating the face and voice of late dictator Suharto. A post on X revealed that the video was generated by AI, but some online critics criticized it as a misuse of AI tools to intimidate and sway voters.
Friday’s agreement states that in responding to AI-generated deepfakes, platforms will “pay attention to context, particularly the protection of educational, documentary, artistic, satirical, and political expression.”
The companies said they are committed to transparency to users about their policies against deceptive AI election content and will work to educate the public on how to avoid being fooled by AI fakes.
Many companies have traditionally put safeguards in place for their generated AI tools that can manipulate images and sounds, while also identifying AI-generated content so that social media users know if what they’re seeing is real. He also said that efforts are being made to label the products. . But most of the proposed solutions have yet to be deployed, and companies face pressure from regulators and others to do more.
That pressure is mounting in the United States, where Congress has yet to pass legislation regulating AI in politics, leaving AI companies largely to govern themselves. In the absence of federal legislation, many states are considering ways to put guardrails around the use of AI in elections and other uses.
The Federal Communications Commission recently acknowledged that AI-generated audio clips from robocalls violate the law, but not audio deepfakes distributed on social media or campaign ads.
Misinformation experts say that while AI deepfakes are particularly concerning for their potential to fly under the radar and influence voters this year, cheaper and simpler forms of misinformation remain a major threat. I’m warning you that there is. The agreement also addressed this, acknowledging that “traditional operations (‘cheapfakes’) may be used for similar purposes.”
Many social media companies already have policies in place to stop deceptive posts about the election process, whether generated by AI or not. For example, Meta will remove false information about “dates, places, times, and methods of voting, registering to vote, participating in the census,” or other false posts intended to prevent someone from participating in the civic engagement. states.
Jeff Allen, a co-founder of the Integrity Institute and a former Facebook data scientist, said the agreement appears to be a “positive step,” but it still requires social media companies to work with others to combat misinformation. He said that he hopes that basic actions will be taken. A content recommendation system that doesn’t prioritize engagement above all else.
In addition to the major platforms that helped broker Friday’s agreement, other signatories include chatbot developers Anthropic and Inflection AI. Eleven Lab is a voice cloning startup. chip designer Arm Holdings; Security companies McAfee and Trend Micro. And Stability AI is known for creating his Stable Diffusion image generation tool.
Notably, another popular AI image generation company, Midjourney, is not included in the agreement. The San Francisco-based startup did not immediately respond to a request for comment Friday.
X’s participation, which was not mentioned in previous announcements about the pending deal, was one of the biggest surprises of Friday’s deal. Since taking over the former Twitter Inc., Musk has drastically reduced his content management team and described himself as a “free speech absolutist.”
But in a statement Friday, X CEO Linda Yaccarino said “all citizens and businesses have a responsibility to protect free and fair elections.”
“X is committed to doing our part, working with our peers to combat the threat of AI, while protecting free speech and maximizing transparency,” she said. .
__
The Associated Press receives support from several private foundations to enhance our explanatory coverage of elections and democracy. Learn more about AP’s Democracy Initiative. AP is solely responsible for all content.
Matt O’Brien and Ali Swenson, Associated Press