Meta executive reveals big tech’s thinking about AI and copyright as legal showdowns loom

Microsoft, headed by CEO Satya Nadella, is being sued by The New York Times, which alleges the tech giant’s powerful AI models used millions of articles for training without permission.

Comments by the chief AI scientist at Meta, Facebook’s owner, have given a glimpse into the thinking of tech giants about copyright and training their powerful artificial intelligence engines.

Yann LeCun wrote on X that “only a small number of book authors make significant money from book sales”.

“This seems to suggest that most books should be freely available for download,” he wrote, before trying to appear more altruistic about ripping off struggling writers.

“The lost revenue for authors would be small and the benefits to society large by comparison.”

Meta chief AI scientist Yann LeCun.

LeCun contradicted Australia’s former competition tsar, Rod Sims, who warned that the rise of artificial intelligence threatened to damage society by abolishing any incentive to produce art and other creative content – as copyright becomes the major battleground for the much-hyped technology

Generative AI – the ability to produce content, from corporate reports to sonnets via simple written prompts – has been described as a wonder technology that will turbocharge productivity.

But AI relies on data – or more precisely other people’s content, such as novels, newspaper articles, poems and other literature – to deliver on its promises. The higher the quality of content, the better AI performs; otherwise it risks becoming a cesspool of misinformation and rubbish.

Artists, musicians and publishers say the tech giants must compensate creators for using their work to train AI engines.

It is an argument that is being tested on both sides of the Atlantic. The New York Times is suing Microsoft and ChatGPT-maker Open AI, alleging that the companies’ powerful AI models used millions of articles for training without permission.

Meanwhile, in the UK stock picture company Getty Images is suing Stability AI over photos used to power the technology company’s Stable Diffusion image generator.

Former Australian Competition and Consumer Commission chair Rod Sims. Picture: Gary Ramage

LeCun wasn’t acting as a rogue agent when he took to X. He was responding to a post from Microsoft senior director Tren Griffin, who wrote: “A survey of 5699 published authors found that in 2022 their median gross pre-tax income from their books was $2000.”

If that wasn’t enough to make a struggling writer weep, Griffin added: “You can generate more revenue than that in less time by mowing lawns in your neighbourhood.”

In other words, it’s OK to steal someone else’s content because they don’t make much money from it. It’s a stupid argument – and a sign that Meta boss Mark Zuckerberg and Microsoft chief executive Satya Nadella need better staff.

After all, would Zuckerberg or Nadella condone one of their executives robbing a small business – or to keep the analogy at scale, a cake stall at a school fete – because it doesn’t make much money?

Would they walk into a newsagent and pinch a newspaper? Or a loaf of bread from a baker (200 years ago, you’d get sent to Australia for such a crime).

But seriously, if big tech bosses think such blatant theft is OK, then it’s no wonder public trust in the tech titans has plummeted. Just because something can be downloaded or is viewable online doesn’t mean you can steal it.

Meta boss Mark Zuckerberg.

Tech companies say that AI is creating a public good, and indeed it has the potential to. But it shouldn’t come at the expense of those who are making genuine, creative content and are already struggling to make a living. After all, the tech giants are making squillions.

OpenAI’s annualised revenue hit $US1.6bn ($2.37bn) in 2023 and is on track to reach $US5bn by the end of 2024, according to American news website Axios.

Microsoft, meanwhile, is charging its enterprise clients $US30 per employee, per month, to access its AI co-pilot. With millions of workers globally subscribed to Microsoft 365, it is set to make the company billions of dollars.

Google is charging a similar price for its AI assistant.

Microsoft chief Satya Nadella meets Prime Minister Anthony Albanese.

The big tech companies argue that AI platforms are expensive to run, given the sheer computing power needed to operate them, not to mention the data storage costs. Indeed, Microsoft is spending $5bn building nine new data centres in Australia to power the explosion in cloud computing from the rapid uptake of AI, while bolstering the nation’s cyber defences.

The thing is, high-quality content is also expensive to produce, and Rod Sims is not alone when he warns of the corrosive effects AI will have on society if the tech giants don’t pay up.

Steve Hasker, chief executive and president of Thomson Reuters – one of the world’s oldest and biggest news and information companies – says publishers and content creators must be compensated fairly.

“We spend a lot of money generating what we think is world-class, independent, fact-based news, and ensuring that there is a viable business model to support that is incredibly important for the functioning of financial markets, of our political system and really society itself,” Hasker says.

News Corporation global chief executive Robert Thomson says AI must not be allowed to become “degenerative AI”.

Robert Thomson, global chief executive of News Corp – owner of this masthead – puts it more simply, saying the technology must not be allowed to be “degenerative AI”.

“There has been much discussion, some of it enlightened, some not so, about the potential impact of generative AI and there is no doubt that it will profoundly affect the media business – candidly, generative AI may pose a challenge to our intellectual property and to the future of journalism,” Thomson said last year.

“As those who have experimented with ChatGPT will be aware, the answers are only as insightful and factual as the source material and are more retrospective than contemporary.”

The good news is that in Australia the federal government has left the door open to force tech giants to pay for content to train their AI engines as part of the media bargaining code.

But as Sims says, the risk the technology poses to copyright stretches beyond news media.

“Once you lose that link to ownership and reward of something you’ve created, then you are really damaging society because you are taking away the incentive of people to create news or write a novel, and so having this technology can completely override copyright and override ownership.”