EXCLUSIVE: Creators' hero unveils ethical tech venture in fight to preserve human creativity
#101 | PLUS: AI warrior questions UK's Sovereign AI policy | German news orgs demand stronger AI regulation |✨AND: It's not just annoying, it's everywhere!
WELCOME TO Charting Gen AI, the global newsletter that keeps you informed on generative AI’s impacts on creators and human-made media, the ethics and behaviour of the AI companies, the future of copyright in the AI era, and the evolving AI policy landscape. Our coverage begins shortly. But first …
LEAD STORY
🎨 AI & CREATIVITY
CREATORS’ HERO Ben Zhao unveiled a new venture this week as part of his mission to preserve human-made media in the era of gen AI. The computer scientist whose University of Chicago lab famously pioneered AI-defeating tools Glaze and Nightshade launched non-profit offshoot ETCH — Ethical Technology and Computing for Humanity — grounded in the belief that tech “should serve humanity and that human creativity is essential and irreplaceable”.
ETCH’s first product is an AI music detector, launched in the same week that Paris-headquartered music platform Deezer revealed the proportion of AI-generated tracks being delivered to it each day had reached 44% of all new uploaded music. More on Deezer’s measures in fighting AI slop in a moment, but first we return to ETCH and the ‘mark’ that its founder continues to make in the pursuit of technologies that benefit creators, and sustain human creativity.
Charting Gen AI caught up with Professor Zhao who last month teased the launch of an AI music detector in his interview for the inaugural Creators Emergency Summit, convened by Charting and co-hosted by international law firm Addleshaw Goddard. In that conversation Zhao said his realisation that music platforms were drowning in AI slop occurred after listening to the radio while on a family vacation in Italy. When he returned home he downloaded the songs but then suspected they were AI generated. “I checked, and sure enough they were. And that was really a wake-up call, understanding how far they’ve come.” Zhao’s SAND Lab — Security, Algorithms, Networking and Data — subsequently found that up to half of all new songs uploaded to platforms such as Spotify are now generated by AI.
FAST, EASY, ACCURATE
The music detector that his team has developed is called Quicksilver. Based on research shared by Deezer, Quicksilver — which is available as a browser plugin and MacOS app — identifies artifacts present in songs generated using Suno or Udio. Charting tested the Quicksilver Chrome plugin on three well-known AI tracks on YouTube: Velvet Sundown’s Dust on the Wind; Xania Monet’s How Was I Supposed to Know?; and Breaking Rust’s Walk My Walk. Quicksilver was fast (it only needs a 30-second sample), easy to use, and accurate:
FREE FOR CREATORS
Speaking exclusively to Charting, Zhao described ETCH as “a natural extension of the research and advocacy work that folks from my lab and I have been actively engaged in”. “I’ve personally worked with so many different creatives across a range of disciplines and modalities that it felt appropriate and necessary to have a non-academic way to engage with them.
“As for the non-profit approach, that was an easy call. Much of the harms, the hype, and questionable decisions around AI are driven by a desperate race to maximise profit. It is the exact opposite of what we set out to do when we first started Glaze. We want to help preserve and protect human creativity, not profit from its adversity.”
While ETCH is non-profit, Zhao — who is president of the venture — said he was “interested in licensing revenue as a way to fund our non-profit work”. “Our tools and work will always be free for individual creatives,” he stressed. “But we also recognise that companies and commercial entities also need our help. A number of companies have reached out to us about licensing, and we are exploring potential relationships.”
DEEZER DATA
Meanwhile Deezer said it was now receiving nearly 75,000 AI-generated tracks each day, 44% of the daily uploads. That’s up from 60,000 AI tracks, 39% of the intake, in January this year (Charting #89). The music streamer — the only platform that explicitly tags AI-generated tracks — said consumption of AI songs was still very low, “between 1% and 3% of the total streams”. Most (85%) of those streams had been detected as fraudulent and were demonetised.
“AI-generated music is now far from a marginal phenomenon and as daily deliveries keep increasing, we hope the whole music ecosystem will join us in taking action to help safeguard artists’ rights and promote transparency for fans,” said Alexis Lanternier, CEO of Deezer.
“Thanks to our technology and the proactive measures we put in place more than a year ago, we have shown that it’s possible to reduce AI-related fraud and payment dilution in streaming to a minimum. Since January, we have made our detection technology available for licensing, and we’re looking forward to seeing industry peers of all kinds join us in the fight for fairness in the age of AI.”
📣 COMMENT: As we said last month, Ben Zhao is a rarity in his field. He fully understands the issues that creators face, and is committed to helping their cause. Why? A passionate belief that human creativity “is the best of us”, and a mission to prevent “a dark world where we’re just consuming regurgitated content from AI slop over and over again for the rest of our lives”. Asked who would find Quicksilver useful he said “anyone who wants to know the source of the music they are listening to, especially those who are interested in listening to and supporting the livelihood and success of human musicians”. Having “first pushed for the development of Quicksilver” because he “needed it” he was “literally its first user”. ETCH is the lab to not only watch, but also engage with as creators look for protections.
▪️Planning has begun on the next Creators Emergency Summit... the event that’s spawning a movement. We want to hear about strategies and solutions that will help publishers and creators defend their businesses and individual livelihoods from AI exploitation. This could be in rights protection and management; content labelling and authentication; systems enabling content licensing, and those forcing greater transparency. It could also be a legal action you’re taking or considering, as well as a campaign to persuade politicians to do the right thing. Do get in touch!
RELATED:
PLUS:
➡ YouTube shares its AI detection tool with studios: Hollywood Reporter
➡ 86% of consumers across seven countries want AI content to be labelled
➡ The music orgs who’re rushing to set standards to protect artists’ rights
➡ MBW predicts how AI will be used to protect musicians’ IP rights
➡ AI movie director dismisses IP concerns over AI saying ‘fuck copyright’
➡ Dataland, the first museum of AI ‘arts’, to open in Los Angeles in June
➡ Instagram’s flash filter is an instant hit, but it changes how people look
➡ Tencent clamps down on AI slop after couple make fortune on WeChat
➡ Runway boss says AI could make 50 movies for price of one blockbuster
ALSO THIS WEEK
🏛️ AI POLICY & REGULATION
THE HEAD of the UK’s recently launched Sovereign AI Fund has been accused of refusing to answer a crucial question on AI and copyright. The fund — a £500 million ($675 million, €576 million) bet on homegrown AI start-ups — is chaired by venture capitalist James Wise, partner at tech investor Balderton Capital. On X ethical training warrior Ed Newton-Rex asked Wise whether the fund would “invest in companies that train AI models on copyrighted work without a licence”:
