Microsoft Urges Congress to Ban AI-Generated Deepfake Fraud

Microsoft is urging Congress to regulate AI-generated deepfakes to mitigate risks of fraud, manipulation, and abuse. Brad Smith, Microsoft's vice chair and president, emphasizes the need for immediate legislative action to safeguard elections and protect vulnerable populations such as seniors and children.

"In light of recent efforts by the tech sector and non-profit organizations, it's clear that our laws must evolve to address deepfake fraud," Smith states in a blog post. He advocates for a comprehensive deepfake fraud statute that will enable law enforcement to prosecute AI-generated scams effectively.

Additionally, Smith calls on lawmakers to update federal and state laws concerning child sexual exploitation and non-consensual intimate imagery to include AI-generated content. Recently, the Senate passed a bill targeting sexually explicit deepfakes, which allows victims to seek damages from perpetrators of non-consensual AI deepfakes. This development follows alarming incidents where middle and high school students created explicit images of peers, and trolls shared harmful AI fakes of public figures.

In light of these challenges, Microsoft has taken steps to enhance safety controls for its AI products after a flaw in its Designer AI image creator was exploited to generate explicit images of celebrities. "The private sector must innovate and implement safeguards to prevent AI misuse," Smith asserts.

While the FCC has banned robocalls using AI-generated voices, the rise of generative AI has made it increasingly easy to produce fake audio, images, and videos—especially relevant as the 2024 presidential election approaches. A recent incident saw Elon Musk share a deepfake video impersonating Vice President Kamala Harris, which may violate X's policies against synthetic media.

To foster transparency and trust in the digital information landscape, Microsoft urges Congress to mandate that AI system providers employ advanced provenance tools to label synthetic content. “This is vital to help the public discern whether content is AI-generated or manipulated,” Smith concludes.