Sam Altman, CEO of OpenAI and the human face of artificial intelligence (AI), recognized this this Wednesday at the Davos forum. His ChatGPT tool, he said, “is going to make a lot of people uncomfortable.” He already does it. Those on TikTok have had access to rare videos in which the big music star, a robotic Taylor Swift, promises free sets of kitchen utensils.
The real Swift never participated in the campaign. These videos, and much worse, are scams generated thanks to the application of AI. This technology allows clones of people to be created without their consent. Sometimes, as she reported at a school in New Jersey, to spread images of naked girls and teachers. In others, as is the case at another Seattle school, students are investigated who have made disturbing images of classmates.
Faced with this issue, which greatly worries celebrities due to the possible exploitation of their image for perverse purposes, the United States Congress has proposed new legislation to combat AI deepfakes, those ultra-fake videos that give the entire appearance of be real thanks to extreme manipulation with that tool. One of the sources of this scam is the dissemination of fake nudes of celebrities, which adds to the serious problem that has arisen online in many institutes.
Despite the popularity of websites and applications designed to generate these deepfakes, there is no legislation in the United States to prohibit the creation of these images by AI. The regulatory proposals, introduced by legislators from both parties, will establish the legal definition of image and voice rights. This means a veto to the use of AI deepfakes to imitate another person, whether alive or dead, without consent.
The draft of this law proclaims that “each individual has the right to property over his or her image and voice.” This document cites various incidents in which several people have seen how they were recently replicated in strange robots thanks to AI. In these examples, special mention is made of recent viral videos in which songs are presented by artificial intelligence in which the voices of pop artists such as Justin Bieber, Bad Bunny, Drake and The Weeknd are imitated.
Another objective is directed against deepfake AI porn. This includes any representation or voice that shows sexual abuse of minors or explicit material. This fits the definition of damage from this standard.
