To help you stay on top of the latest news, our AI practice group has compiled a roundup of the developments we are following.
- The European Union’s AI Act, containing sweeping regulation of emerging artificial intelligence technologies, entered into force this week. According to a press release from the European Commission, the legislation is the world’s first comprehensive regulation on AI, designed to ensure AI developed and used in Europe is trustworthy and protects fundamental rights. The Act introduces a risk-based approach, categorizing AI systems into minimal, specific transparency, high and unacceptable risk levels, with corresponding obligations and restrictions. Those with minimal or with a specific transparency risk have relatively lenient restrictions, including disclosure that users are interacting with an AI system in the latter case. However, AI systems designated as “high risk,” such as AI systems used for employee recruitment, loan applications or autonomous robots, must comply with strict requirements, including “risk-mitigation systems, high quality of data sets, logging of activity, detailed documentation, clear user information, human oversight, and a high level of robustness, accuracy, and cybersecurity.” Lastly, those designated as an “unacceptable risk,” including AI systems that “manipulate human behavior to circumvent users’ free will,” and “systems that allow ‘social scoring’ by governments or companies,” will be banned. The AI Act also establishes rules for general-purpose AI models and outlines enforcement mechanisms, including fines of up to 7% of global annual turnover for violations of bans, up to 3% for violations of other obligations, and up to 1.5% for supply incorrect information. While most rules will apply beginning in August 2026, some provisions, such as prohibitions on high-risk systems, will take effect after six months. In particular, EU member states have one year to designate authorities within their respective nations to oversee the implementation of the AI Act’s rules, while the newly created European Artificial Intelligence Board will ensure uniform application across the EU. This Roundup will continue to monitor this major development in AI policy as the Act is implemented over the next few years.
- A bipartisan group of U.S. senators have introduced a bill to protect individuals’ voices and likenesses from AI-generated replicas. The Nurture Originals, Foster Art, and Keep Entertainment Safe Act of 2024, or “NO FAKES” Act, was introduced this week by Senators Chris Coons (D-DE), Marsha Blackburn (R-TN), Amy Klobuchar (D-MN), and Thom Tillis (R-NC), and will “hold individuals or companies liable for damages for producing, hosting, or sharing a digital replica of an individual performing in an audiovisual work, image, or sound recording that the individual never actually appeared in or otherwise approved – including digital replicas created by generative artificial intelligence.” The bill aims to create a national standard for protecting individuals’ rights to their own voice and likeness, with exceptions for First Amendment protections such as documentaries, biographies and parody. It has garnered support from various entertainment industry organizations, including the actors’ union SAG-AFTRA, publishers such as the Recording Industry Association of America and the Motion Picture Association, and major technology companies such as OpenAI and IBM.
- Reuters reports on the ongoing copyright infringement lawsuit brought by music labels Universal Music Group, Warner Music Group and Sony Music Group against AI-generated music startups Suno and Udio. We covered the original suit in a previous AI roundup. The AI companies have responded to the lawsuits in federal court, arguing that their use of copyrighted sound recordings to train their AI systems qualifies as fair use under U.S. copyright law. Suno and Udio, which have raised millions in funding for their text-to-music AI systems, claim the lawsuits are attempts to stifle independent competition. The startups compare the labels’ protests to past concerns about new technologies in the music industry, such as vinyl records and synthesizers. However, the Recording Industry Association of America maintains that the AI companies are unfairly exploiting artists’ work. These lawsuits are among the first to target music-generating AI, following similar cases in other creative industries, including The New York Times’ ongoing suit against OpenAI, also alleging copyright infringement of its articles.
- The U.S. Copyright Office released the first portion of its report on AI’s implications for copyright, specifically concerning “digital replicas,” defined as “videos, images, or audio recordings that have been digitally created or manipulated to realistically but falsely depict an individual.” According to Engadget, the report concludes that state-level regulations are a “patchwork of protections” that are insufficient to protect creators, and existing federal laws are too narrow to address the harms digital replicas inflict. The Copyright Office recommends new federal legislation to address the challenges posed by AI-generated digital replicas, popularly known as “deepfakes”, emphasizing the need for prompt action due to the speed, precision and scale of modern digital replicas. The report suggests implementing safe harbor provisions to encourage online service providers to quickly remove unauthorized digital replicas and advocates for laws that protect all individuals, not just celebrities or others whose livelihood depends on their likenesses and voices. This publication of the Copyright Office’s report coincides with recent legislative action in Congress, including the passage of the DEFIANCE Act and the introduction of the NO FAKES Act, both aimed at providing legal recourse for victims of unauthorized digital replications.
- OpenAI is rolling out its delayed voice-assistant feature to a select group of ChatGPT Plus users, according to Bloomberg. The company is making the voice feature available after previously postponing the release to address safety concerns. The new feature offers four preset voices (though not including the “Sky” voice which ignited controversy for its similarity to Hollywood actress Scarlett Johansson) and includes additional safeguards to prevent impersonation and copyright infringement, such as filters to detect and refuse requests for generating copyrighted audio content. OpenAI emphasizes that this gradual rollout allows for close monitoring and continuous improvement based on real-world feedback. While the voice assistant was initially announced as part of the GPT-4o update in May, some promised capabilities, such as video and screen-sharing features, are still in development with no specific launch date. The company plans to expand the voice feature to all ChatGPT Plus subscribers in the fall of 2024.
[View source.]