Tech giants including Meta, Microsoft and TikTok signed a pledge on Friday to crack down on AI content aimed at misleading voters ahead of this year's key elections around the world.
Under the agreement, companies including Google and OpenAI pledge to develop ways to identify, label, and control images, videos, and audio generated by AI for the purpose of misleading voters.
“I think it takes all the stakeholders, from the source of generation to the consumption by the users who are actually involved. That's why having all 20 companies sign up for this is so impactful. “We believe that,” said Nick Clegg, Meta's global president. he told AFP.
The 20 signatories to the agreement, announced on the sidelines of the Munich Security Conference in Germany, include X, formerly known as Twitter, as well as Snap, Adobe, LinkedIn, Amazon and IBM. was.
Under the agreement, AI-generated content can be watermarked and tagged with source metadata, but the signatories note that “such solutions have limitations.” is recognized.
Advertisement – SCROLL TO CONTINUE
The tech companies said they would also help develop ways to “detect and address” fraudulent election materials on their platforms.
For example, such content can be annotated to make it clear that it is AI-generated.
Meta, Google, and OpenAI have already agreed to use a common watermarking standard to tag images produced by AI applications such as OpenAI's ChatGPT, Microsoft's Copilot, and Google's Gemini (formerly known as Bard). .
Advertisement – SCROLL TO CONTINUE
The pledge comes as big tech companies are under intense pressure over concerns that their AI-powered applications could be misused during a critical election year.
European Commission Vice-President for Values and Transparency Vera Jurova, who attended the Munich presentation, said tech companies had “agreed that technology can pose risks to democracy”. He said he was happy.
But the government “cannot blame everything on big technology companies,” Jourois said.
Advertisement – SCROLL TO CONTINUE
As the EU prepares for European Parliament elections in June, he said, “We have some things to do.”
Recent examples of convincing AI deepfakes only heighten concerns about easily accessible technology.
Last month, robocalls impersonating U.S. President Joe Biden went out to tens of thousands of voters, urging them not to vote in the New Hampshire primary.
In Pakistan, former Prime Minister Imran Khan's party used AI to create a speech by its jailed leader.
sea/hmn/rl
