\u201cThe current uncertainty around intellectual property (IP) is hindering innovation and undermining our broader ambitions for AI, as well as the growth of our creative industries,\u201d the plan states.<\/p>\n
\u201cThis has gone on too long and needs to be urgently resolved. The EU has moved forward with an approach that is designed to support AI innovation while also enabling rights holders to have control over the use of content they produce. The UK is falling behind.\u201d<\/p>\n That has drawn criticism from some observers, who fear that AI-friendly copyright reforms could undermine the value of intellectual property.<\/p>\n \u201cThe recommendation to reform copyright law to favor AI companies is terrible news for the creative industries,\u201d said Ed Newton-Rex<\/strong>, founder and CEO of Fairly Trained<\/strong>, in a post<\/a> on LinkedIn.<\/p>\n \u201cThe recommendation to reform copyright law to favor AI companies is terrible news for the creative industries.”<\/p>\n Ed Newton-Rex, Fairly Trained<\/span><\/p><\/blockquote>\n Fairly Trained<\/a> is a non-profit that certifies AI models that have been trained ethically (i.e., without the unauthorized use of copyrighted content).<\/p>\n \u201cWe could get all the benefits of the Action Plan, and be a leader in AI, without upending copyright law,\u201d added Newton-Rex, who in 2023 resigned<\/a> as VP of Audio at Stability AI<\/strong> over the company\u2019s insistence that using copyrighted materials without permission to train AI amounts to \u201cfair use\u201d under US copyright law.<\/p>\n Dan Conway<\/strong>, CEO of the Publishers Association<\/strong>, which represents book and periodical publishers, warned that this reform could lead to \u201ca US tech-led heist of UK copyrighted works.\u201d<\/p>\n In a LinkedIn post<\/a> on Monday, Conway noted Parliament is currently running a consultation on AI and copyright, and said the Publishers Association \u201cwill be seeking assurances from Ministers that all options \u2013 including enforcing the existing gold standard copyright regime with transparency provisions \u2013 remain under full consideration and proper scrutiny.\u201d<\/p>\n At this point, the details of how the UK plans to reform its copyright laws are unknown, but the reference to the European Union offers some guidance.<\/p>\n The EU\u2019s AI Act<\/a>, which Brussels has promoted as the world\u2019s first comprehensive law governing artificial intelligence, creates an \u201copt-out\u201d regime for AI data mining. AI developers can scrape the internet for copyrighted materials to train their models, unless a copyright holder expressly states that they object.<\/p>\n “The UK can achieve public service reform and seize all the growth opportunities associated with AI without facilitating a US tech-led heist of UK copyrighted works.”<\/p>\n Dan Conway, Publishers Association<\/span><\/p><\/blockquote>\n That model has irritated many music rights holders, who are pushing for an \u201copt-in\u201d model, under which AI developers can\u2019t use copyrighted content to train AI unless they get permission from rights holders. They argue that the opt-out model unfairly puts the onus on copyright owners to enforce their rights.<\/p>\n Nonetheless, some music majors took advantage of the EU law to clarify that they don\u2019t consent to their copyrighted works being used for AI without prior authorization.<\/p>\n In May of last year, Sony Music Group<\/strong> sent a letter<\/a> to 700 AI developers and music streaming services declaring it\u2019s opting out of having its materials used to train AI. A month later, Warner Music Group<\/strong> sent a letter to tech companies notifying them they need permission to use its content<\/a>.<\/p>\n Another point in the proposed AI plan aims to create a \u201ccopyright-cleared\u201d library of media content for AI developers. According to the Action Plan, the so-called \u201cBritish media asset training data set\u201d would potentially be sourced from institutions such as the National Archives, the Natural History Museum, the British Library, and the BBC.<\/p>\n The UK\u2019s AI plan comes amid growing competition among nations for a key role in the development of artificial intelligence, and ongoing disputes between rights holders and tech companies over the use of copyrighted works to train AI models.<\/p>\n In the US, copyright owners have filed numerous lawsuits against AI developers, alleging that the tech companies used copyrighted materials without permission to train their AI.<\/p>\n Among the most prominent of these are two lawsuits<\/a> filed by divisions of Sony Music Group<\/strong>, Universal Music Group<\/strong>, and Warner Music Group<\/strong> against Suno<\/strong> and Udio<\/strong>, generative AI services that enable users to create entire songs from text prompts quickly.<\/p>\n The music majors allege Suno and Udio trained their models on their works, pointing to similarities between existing copyrighted songs and content created by the AI music-making platforms.<\/p>\n For their part, Suno and Udio have all but admitted<\/a> that they likely used copyrighted material, but argue that this falls under \u201cfair use.\u201d<\/p>\n![\"Ed](\"https:\/\/www.musicbusinessworldwide.com\/files\/2024\/01\/ed-newton-rex-80x44.jpg "\\"Ed")
<\/figure>