{"id":14271,"date":"2024-01-18T14:21:53","date_gmt":"2024-01-18T14:21:53","guid":{"rendered":"http:\/\/TheNextWeb=1403080"},"modified":"2024-01-18T14:21:53","modified_gmt":"2024-01-18T14:21:53","slug":"certificates-for-copyright-compliant-ai-are-here-but-chatgpt-wont-get-one","status":"publish","type":"post","link":"https:\/\/www.londonchiropracter.com\/?p=14271","title":{"rendered":"Certificates for copyright-compliant AI are here \u2014 but ChatGPT won\u2019t get one"},"content":{"rendered":"\n

A new certificate for copyright-compliant AI<\/a> aims to protect creators \u2014 but don\u2019t expect ChatGPT to get the stamp of approval.<\/span><\/p>\n

Named <\/span>Fairly Trained<\/span><\/a>, the initiative arrives amid a brewing backlash against generative AI companies. Many of their tools \u2014 from OpenAI\u2019s<\/a> chatbots to Stability AI\u2019s art generators \u2014 are trained on copyrighted content that\u2019s scraped from the web. Inspired by this data, the systems then deliver endless creations in response to prompts. Frequently, the outputs are clear derivations<\/a> of their source material.<\/span><\/p>\n

The practice has enraged creators and copyright-holders. They argue that their work is stolen and remixed without their consent and compensation.<\/span> It\u2019s hardly a controversial claim \u2014 GenAI leaders <\/span>have admitted<\/span><\/a> to the practice. <\/span><\/p>\n

To justify the process, companies point to the \u201cfair use\u201d doctrine, which can allow<\/span> transformative and socially beneficial use of copyrighted content.<\/span><\/p>\n

\n
<\/p>\n

TNW Conference 2024 – Speakers announced!<\/p>\n

Meet the powerhouse experts that will take the stage on June 20 & 21 in Amsterdam and save your seat today.<\/p>\n

<\/a><\/div>\n<\/div>\n

That argument has sparked opprobrium. One of the most prominent critics is <\/span>Fairly Trained\u2019s CEO, Ed Newton-Rex. A musician and computational creativity pioneer, Newton-Rex attracted headlines<\/a> in December after quitting his role at Stability AI over the startup\u2019s<\/a> use of copyrighted content. The 36-year-old raised concerns that the company was \u201cexploiting creators.\u201d <\/span><\/p>\n

Fairly Trained is his attempt to foster an alternative model. By certifying companies that get a license for their training data, the non-profit wants to create a fairer world for human creators. <\/span><\/p>\n

The verification shows which companies consider creator consent to be important \u2014 and which ones don\u2019t. Consumers can then make an informed decision on their use of GenAI.<\/span><\/p>\n

<\/p>\n

\n
\n

\u201cGenAI poses an existential threat to creative industries.<\/p>\n<\/p><\/div>\n<\/blockquote>\n

<\/span><\/p>\n

The idea emerged after Newton-Rex resigned from Stability AI. In the debates triggered by his departure, he realised that licensed GenAI tools needed more exposure.<\/span><\/p>\n

\u201cBoth for ethical and legal reasons, there are a lot of people and companies who would rather use generative AI models that are that are trained on licenced data,\u201d Newton-Rex told TNW. \u201cYou\u2019ve got a bunch of people who want to use licenced models and you\u2019ve got a bunch of people who are providing those. I didn\u2019t see any way of being able to tell them apart.\u201d <\/span><\/p>\n

Fairly Trained provides one way to differentiate them. The inaugural batch of certificates was awarded to nine GenAI organisations: <\/span> <\/span>Beatoven.ai<\/span><\/a>,<\/span> Boomy<\/span><\/a>,<\/span> Bria AI<\/span><\/a>,<\/span> Endel<\/span><\/a>,<\/span> LifeScore<\/span><\/a>,<\/span> Rightsify<\/span><\/a>,<\/span> Somms.AI<\/span><\/a>,<\/span> Soundful<\/span><\/a>, and<\/span> Tuney<\/span><\/a>. While the majority are music makers, image generators are also represented and other media formats are \u201con the way,\u201d Newton-Rex said. <\/span><\/p>\n

There is, however, one big gaping in the modalities: text.<\/span> Newton Rex isn\u2019t aware of any major text generation model that could currently get certified.<\/span><\/p>\n

\u201cI don\u2019t know any that would pass the bar, because every large language model that I\u2019ve come across has been trained on a huge amount of copyrighted work,\u201d he said.<\/span><\/p>\n

\"Photo
Credit: Ed Newton-Rex<\/figcaption>
<\/i><\/a>Newton-Rex was previously Stability AI\u2019s vice president for audio. He also founded the music generator Jukedeck, which Bytedance acquired in 2019.<\/span><\/figcaption>