The FCC is proposing to outlaw the use of AI-generated voices in robocalls, making it easier to charge operators of these fraudulent calls. The use of voice cloning technology, such as using President Biden’s voice, would be deemed fundamentally illegal. This move aims to give State Attorneys General offices new tools to crack down on these scams and protect consumers. The FCC is taking steps to recognize this emerging technology as illegal under existing law.
Similarly, IBM researchers have discovered a method for scammers to hijack voice calls using generative AI tools, allowing them to impersonate someone’s voice and manipulate ongoing conversations to steal funds and sensitive information. The researchers call this threat “audio-jacking,” which involves using voice clones to replace certain phrases with a victim’s cloned voice. This poses a significant risk to financial institutions and others who rely on phone calls for identity verification. The report suggests paraphrasing and repeating what was said to verify accuracy and trip up the chatbots used in these scams.
From Politico, Conservatives, tech lobbyists, and GOP lawmakers are opposing President Joe Biden’s executive order on artificial intelligence (AI), which uses emergency powers to compel tech companies to provide information to the Commerce Department about advanced AI projects. Critics argue that using the Defense Production Act to regulate AI is an executive overreach and not what the law intended.
I was also taken with this dual use of copyright law. Meta, formerly known as Facebook, used copyright law to protect its Llama AI model while arguing against the same protections for others. The company believes that online content used for free to build AI models falls under fair use, except for its content. Meta’s attempt to have a leaked version of Llama removed from GitHub failed, as the model’s specifications were deemed not copyrightable. Other Big Tech companies, including Microsoft and Apple, argue for copyright protection for their AI models’ outputs while claiming that the copyrighted inputs used to train the models cannot be protected.
The pushback highlights the broader political disagreements over the regulation of private enterprise, the potential impact on AI safety standards, and the technology’s growth. I think the executive order is smart and would challenge lawmakers to propose some bills of their own if they disagree. Because companies certainly will use the laws on the books however they see fit.
Robocalls are a scourge, and AI will make the entire thing worse. I’m encouraged to see some action.