A federal judge has halted the implementation of a new California law aimed at combating deepfake videos in elections. The law, which allows individuals to sue for damages over election-related deepfakes, was put on hold by US District Judge John A. Mendez following a preliminary injunction.
Judge Mendez expressed concerns about the risks posed by artificial intelligence and deepfakes but ruled that the law likely infringes on the First Amendment. He critiqued the legislation, stating that it acted more like a blunt tool than a precise instrument, potentially stifling humour and impeding the free exchange of ideas crucial for American democratic discourse.
The legislation was signed into effect by Governor Gavin Newsom last month, with two other bills also signed simultaneously to address the use of artificial intelligence in creating deceptive images or videos in political advertising leading up to the 2024 election. These laws are considered among the strictest in the country.
In response to the injunction, Izzy Gardon, a spokesperson for Governor Newsom, emphasised that these laws are essential in safeguarding democracy and protecting freedom of speech. However, a lawyer representing YouTuber Christopher Kohls, who challenged the law, welcomed the ruling, describing it as a validation of First Amendment protections in the face of advancing technologies.
First Amendment experts and critics of the law had previously raised concerns about its constitutionality and potential governmental overreach. They argued that existing defamation laws already address harmful speech without the need for additional categories of speech created outside constitutional boundaries.
The decision to block the California law is seen as a significant development in the ongoing debate over the regulation of deepfakes and the balance between protecting against misinformation and upholding free speech rights. The outcome of this legal battle could have broader implications for future attempts to legislate in this complex and evolving technological landscape.