•    "Our colleague, Senator Hagerty, drove this point home at our hearing on AI in political speech. As he explained, the definitions in these bills are nebulous, at best, and overly censorious if they’re applied most cynically. They could wind up barring all manner of photos and videos as long as the ill-defined ‘reasonable person’ could deduce an alternative meaning from the content. Needless to say, if we put up a dozen political ads before this Committee, our colleagues would differ on which ones were intentionally misleading. The core question we’re facing is whether or not politicians should have another tool to take down speech they don’t like. But if the amendment before us extends this authority to unpaid political speech, then we’re also talking about an extension of speech regulation that has not happened in the 50 years of our modern campaign finance regime. Now, I also have concerns about the disclaimer provisions and their application. Our political disclaimer regime has for its entire history served a singular purpose: to help voters understand who is paying for or endorsing an advertisement. It has never been applied to political advertisements as a content regulation tool. And on this bill, the problem again lies in exceptions to the definitions. Let’s remember: As of today, Congress has not even settled on a definition for AI, let alone a definition of ‘deepfake’." Source: Sen. Mitch Mcconnel (Republican, Kentucky)

•    "On the other hand, we urge the Committee to reject the approach taken in S.2770, the Protect Elections from Deceptive AI Act. S.2770 would risk infringing on First Amendment-protected speech while also undermining innovative technology that has increased the accessibility of election-related information to voters. Instead of protecting voters from false and deceptive information, S.2770 could give politicians a tool to target accurate content they simply don’t like. The bill also risks chilling and censoring speech disseminated by everyday Americans, journalists, and small businesses. This legislation, in its current form, is also problematic as it relates to liability for intermediaries such as platforms and publishers of user-generated content." Source: Paul Nicholas Lekas, Senior Vice President and Head of Global Public Policy & Government Affairs at Software & Information Industry Association

•    "This bill restricts far more speech than necessary in two other respects, each of which has proven fatal to other laws aimed at restricting false election-related speech. First, it regulates expression that poses little risk of harm: S. 2770 prohibits anyone from disseminating materially deceptive GAI content at all, treating communication between family and friends the same as mass-distributed expression by campaigns.14 Second, it regulates all expression about a candidate from the moment their pursuit of federal office is announced—a hallmark of overbreadth in electoral speech regulations.15 The sheer breadth of S. 2770’s prohibitions are enough to render it obviously unconstitutional." Source: Coalition Letter: American Civil Liberties Union California Policy Center Competitive Enterprise Institute Electronic Frontier Foundation Foundation for Individual Rights and Expression TechFreedom