Big Tech fights back on fair use | Google AI Overviews could prove 'devastating'
#7 | Plus: EU set to clash with AIs | AI search engine is accused of 'plagiarism' | Kutcher rows back on AI remarks | Apple treads cautiously on AI integration
Welcome to this week’s newsletter charting developments in generative AI and the impacts these are having on human-made media.
🏛️ REGULATION & LEGISLATION
“WHAT DO WE WANT? The right to rip off human-made media! When do we want it? NOW!” That’s the battle cry of Big Tech lobby group the Chamber of Progress as it campaigns to preserve the ‘fair use’ defence used by AIs when faced with legal action over copyright infringement. The group — which is backed by Google, Meta and Apple — said its campaign showcased “how AI lowers barriers for producing art” and defended “the longstanding legal principle of fair use under copyright law”. 🔗 Chamber of Progress statement; Generate + Create
🌟 KEY TAKEAWAY: While fair use is a US legal doctrine the debate over the extent to which the defence can be used to justify the training of AI large language models (LLMs) on copyright protected content is global. And the outcome of cases such as The New York Times versus Open AI and Microsoft — which hinges on the definition of ‘transformative’ fair use — is bound to have ramifications beyond US borders. Bizarrely, this advocacy group says it’s campaigning for “technology’s progressive future”. What’s progressive about misappropriating human-made media?
THE EU is on collision course with the AIs over data used to train their LLMs. Reuters said the EU’s newly established AI Office would in early 2025 release a template it expects AIs to follow when revealing how their models were trained. Under Europe’s AI Act organisations deploying general purpose AI models will need to provide “detailed summaries” but the AIs are likely to object, saying this would force them to reveal trade secrets that need to be protected. 🔗 Reuters
ALSO:
➡ Privacy group files complaints to stop Meta training AI on Facebook.
➡ ChatGPT maker OpenAI ‘expands international lobbying team’: FT.
🚨 NEWS MEDIA
A MAJOR INVESTIGATION into Google’s AI-generated summaries suggested they could have a “devastating” impact on news publishers. Press Gazette and a consortium of publishers commissioned SEO agency Authoritas to analyse the impact of AI Overviews — currently being tested in the US — on the search visibility of news orgs. The study found that AI summaries were being widely used on news-related search terms, forcing publishers’ links down search results. Press Gazette editor-in-chief Dominic Ponsford said: “There are still a huge number of unknowns around AI Overviews but our research suggests that if they were returned for all search queries (which could well be the case in future) the impact on publisher traffic from Google could be devastating.” 🔗 Press Gazette
🌟 KEY TAKEAWAY: Google claims links within AI Overviews receive more clicks than within traditional search results. Press Gazette said it was “reluctant to make specific predictions about likely traffic drops” but based on the estimated impact of falling from top result to fifth position a publisher could see an 85% fall “if the result was replaced with an overview for all readers”. Ponsford hopes the Digital Markets Unit will investigate after the UK’s election — before AI Overviews get a UK launch.
PERPLEXITY WAS ACCUSED of “plagiarism” after a Forbes investigation appeared in the AI-powered search engine’s curated content section. Forbes tech editor John Paczkowski said Perplexity had ripped off “most of our reporting” on former Google chief Eric Schmidt’s secretive drone venture prompting Perplexity CEO Aravind Srinivas to admit the curated section had “rough edges”. That drew an angry response on X from Paczkowski who said it amounted to “little more than plagiarism”. Forbes editor Randall Lane said the incident provided “the perfect case study for this critical moment”. 🔗 Paczkowski on X; Lane’s editorial
🌟 KEY TAKEAWAY: Srinivas is on a mission to disrupt Google. In the process he might be helping to destroy journalism. That charge has been levelled at other leading AIs which have trained their LLMs on intellectual property belonging to human-made media companies without permission to create products that for some (perhaps most) consumers will be substitutive. The courts will have to decide if this is fair use. In the meantime, all journalism can do is call out bad behaviour. Loudly.
ALSO:
➡ News site BNN Breaking ‘fed defamatory AI-generated story to MSN’.
➡ NewsBreak used AI tools to republish local news with fictitious bylines.
➡ And here’s Hoodline, another ‘news org’ using AI for fake journalism.
➡ AI news summariser Particle secures content partnership with Reuters.
🎬 TV, FILM
HOLLYWOOD STAR and tech investor Ashton Kutcher prompted fury after hinting that OpenAI’s generative video tool Sora was the future of TV and filmmaking. “Why would you go out and shoot an establishing shot of a house in a television show when you could just create the establishing shot for $100? To go out and shoot it would cost you thousands of dollars,” said Kutcher. Hollywood writers accused Kutcher of talking up tech and talking down the value of human creativity. Kutcher later rowed back his remarks, saying AI wouldn’t replace the film industry “but we need to be prepared”. 🔗 Variety; NME; Kutcher on X
ALSO:
➡ An AI-generated music video is booed at a French animation festival.
➡ EBU sets out vision for the use of generative AI in public service media.
🤖 TECH WATCH
APPLE’S CAUTIOUS APPROACH to generative AI — it’s calling it Apple Intelligence — was praised by analysts and rewarded by the markets. Shares in the hi-tech soared, briefly making it the world’s most valuable company before its market cap slipped just behind arch-rival Microsoft’s. As we reported in our Quick Take, Apple is integrating gen AI across iPhones, iPads, iMacs and Macbooks in a way that seeks to keep data secure by performing operations on devices or within a private cloud. Users will be asked for permission before Siri asks ChatGPT a question. Apple software engineering chief Craig Federighi stressed Apple is being “very careful” with generative AI: “We’re not taking this teenager and telling it to go fly an airplane.” 🔗 Apple Intelligence; Mashable
⏱️ OUR QUICK TAKE: AI? Apple (finally) has an app for that
ALSO:
➡ Why Google’s AI Overviews are STILL recommending glue for pizzas.
➡ Elon Musk drops lawsuit against OpenAI without explaining why.
➡ OpenAI is reportedly on track to double yearly revenues to $3.4 billion.
➡ Microsoft backs down on AI screenshot feature Recall for Copilot PCs.
📣 MARKETING
MONDELĒZ CONSUMER EXPERIENCES chief Jonathan Halvorson told an AI summit that brand marketers faced a 2001 moment “all over again”. Back then brands had to decide whether to jump online; now they were having to decide on the use of AI. “Those who don’t move will be left behind,” he said. Scott Goodson, CEO of creative agency StrawberryFrog, wrote an Inc. op-ed saying: “AI will be a big development for the creative industry … it isn’t an extinction event, but creative businesses, marketing, advertising, and design agencies will need to reorganise and deploy their talent differently.” 🔗 The Drum; Inc.
ALSO:
➡ Gen AI tools will “help you to zig when everyone else is zagging”.
➡ AI ‘will never replicate the spontaneity of Mad Men’s Don Draper’.
🎨 ART & DESIGN
Cara, the social network for artists that filters out AI-generated art, has leapt from 40,000 members to nearly 1 million within a month following Meta’s confirmation it will use images posted to Instagram for AI training. Jingna Zhang launched Cara in 2023 as a reaction to what she believes is the unethical treatment of artists’ copyrighted works. She told WIRED: “Automation, generative AI in the workplace, it’s coming for more than just designers and artists. Lots of white-collar jobs will be automated away. This is not just about one group of people. It’s going to touch people all across our society.” 🔗 WIRED
ALSO:
➡ We’re not going to train gen AI on your artworks, Adobe tells artists.
➡ Futurist: gen AI will enable “new art forms, genres and movements”.
➡ Photographers and artists are using Immersity AI to turn 2D to 3D.
💬 QUOTES OF NOTE
“AI is inherently not about facts and accuracy. You’ll see that in the tiny disclaimer at the bottom of ChatGPT or any of those tools. So for a profession that writes words that are meant to be factual, all of a sudden you’re competing in the marketplace — essentially, the marketplace of information — with all these words that sound plausible, look plausible and have no relationship to accuracy.” — Julia Angwin, writer and CEO of news studio Proof, talking to The Journalist’s Resource
“I actually think one of the reactions to AI-generative music will be a resurgence of really raw, very immediate rough-sounding music, because people will want and crave that.” — Ilā Kamalagharan, artist, producer and creative director speaking at PRS Explores
🤔 AND FINALLY …
US judge Kevin Newsom used ChatGPT to help him reach a conclusion ... and is now urging other judges to give the chatbot a go. Newsom turned to generative AI to help him consider the definition of “landscaping”. While accepting that many in the legal profession would “reflexively condemn” the use of ChapGTP as “heresy” he felt “ChatGPT might have something useful to say about the common, everyday meaning of the words and phrases used in legal texts”. Ralph Losey, a US lawyer who focuses on tech and AI, said: “This is the first case of its kind and deserves careful study by lawyers and judges all over the world.”