The internet is about to get much worse

Greg Marston, a British voice actor, recently discovered an AI-generated clone of his voice called "Connor" online. The clone was trained on a recording Marston had made in 2003, uttering things he had never said. At the time, Marston had recorded a session for IBM and signed a release form allowing the recording to be used in various ways. However, he did not anticipate that IBM would use his voice for anything other than the exact utterances he had recorded. With the help of artificial intelligence, IBM was able to sell Marston's decades-old sample to websites that are using it to create a synthetic voice capable of saying anything. Marston was shocked to find his voice being used on the Wimbledon website during the tennis tournament. IBM has acknowledged Marston's concerns and is in direct discussions with him regarding the matter.

Marston's situation highlights the growing concern among creators about the erosion of trust in the digital space. People are realizing that their contributions to the public domain can be taken, monetized, and potentially used against them. This erosion of trust could lead to a proliferation of untrustworthy content in our digital public spaces. Artists are already deleting their work from platforms like Twitter after the company announced its intention to use platform data to train AI systems. Hollywood writers and actors are also on strike to protect their work from being replaced by AI systems. News outlets like The New York Times and CNN have taken measures to prevent AI chatbots from scraping their content by adding files to their websites.

Creators are now suing AI companies, claiming that their copyrighted works are being used in training data. OpenAI has argued in a separate case that the use of copyrighted data for training AI systems falls under the "fair use" provision of copyright law. While creators fight for control over their work, dubious AI-generated content is flooding the public sphere. NewsGuard has identified 475 AI-generated news and information websites in 14 languages. AI-generated music is saturating streaming platforms and generating royalties for scammers. AI-generated books, including potentially dangerous guides like a mushroom foraging guide, are prevalent on Amazon. Amazon now requires authors who self-publish on its Kindle platform to declare if they are using AI.

This situation reflects the tragedy of the commons, where a common resource is exploited for individual profit. The internet also has its commons, such as Wikipedia and Reddit forums, where volunteers share knowledge in good faith and work to keep bad actors at bay. However, these commons are now being overexploited by tech companies seeking to feed human wisdom, expertise, humor, anecdotes, and advice into their for-profit AI systems. Volunteers who contribute to Wikipedia trusted that their work would be used according to the site's terms, which require attribution. Now, some contributors are debating whether they have legal recourse against chatbots that use their content without proper citation. Regulators, including the European Union, are considering restrictions on AI to ensure transparency and protect copyrighted data used in training AI systems.

Transparency alone is not enough to address the power imbalance between those whose data is exploited and the companies profiting from it. Tim Friedlander, founder and president of the National Association of Voice Actors, has called for ethical standards in the AI industry, emphasizing the need for consent, control, and compensation for actors. These principles should apply to everyone, whether they are professional actors or simply posting on social media. Meaningful consent should not be reduced to hard-to-find opt-out buttons. Determining fair compensation is challenging, especially since most AI bots are currently free services. However, as the AI industry monetizes these systems, there will be a reckoning with those whose works fueled the profits. For individuals like Marston, their livelihoods are at stake, and seeking compensation becomes necessary.

Even for those not directly threatened by AI, the lack of protections against data exploitation raises concerns for anyone considering creative endeavors in the public domain. Without safeguards against AI data overgrazing, the act of creating may feel futile. This would be a significant loss for society as a whole.