Tech firms say laws to protect us from bad AI will limit ‘innovation’. Well, good | John Naughton

Tech firms say laws to protect us from bad AI will limit ‘innovation’. Well, good

John Naughton

For too long, the industry has escaped legal liability in the pursuit of its own interests – and the EU has had enough

A protest against facial recognition software in central London. Police use of live facial recognition is one of the ‘high-risk’ uses targeted by the EU’s AI Act.

Way back in May 2014, the European court of justice issued a landmark ruling that European citizens had the right to petition search engines to remove search results that linked to material that had been posted lawfully on third-party websites. This was popularly but misleadingly described as the “right to be forgotten”; it was really a right to have certain published material about the complainant delisted by search engines, of which Google was by far the most dominant. Or, to put it crudely, a right not to be found by Google.

On the morning the ruling was released, I had a phone call from a relatively senior Google employee whom I happened to know. It was clear from his call that the company had been ambushed by the ruling – its expensive legal team had plainly not expected it. But it was also clear that his US bosses were incensed by the effrontery of a mere European institution in issuing such a verdict. And when I mildly indicated that I regarded it as a reasonable judgment, I was treated to an energetic tirade, the gist of which was that the trouble with Europeans is that they’re “hostile to innovation”. At which point the conversation ended and I never heard from him again.

What brings this to mind is the tech companies’ reaction to a draft EU bill published last month that, when it becomes law in about two years’ time, will make it possible for people who have been harmed by software to sue the companies that produce and deploy it. The new bill, called the AI Liability Directive, will complement the EU’s AI Act, which is set to become EU law around the same time. The aim of these laws is to prevent tech companies from releasing dangerous systems, for example: algorithms that boost misinformation and target children with harmful content; facial recognition systems that are often discriminatory; predictive AI systems used to approve or reject loans or to guide local policing strategies and so on that are less accurate for minorities. In other words, technologies that are currently almost entirely unregulated.

The AI Act mandates extra checks for “high-risk” uses of AI that have the most potential to harm people, particularly in areas such as policing, recruitment and healthcare. The new liability bill, says MIT’s Technology Review journal, “would give people and companies the right to sue for damages after being harmed by an AI system. The goal is to hold developers, producers and users of the technologies accountable and require them to explain how their AI systems were built and trained. Tech companies that fail to follow the rules risk EU-wide class actions.”

Right on cue, up pops the Computer & Communications Industry Association (CCIA), the lobbying outfit that represents tech companies in Brussels. Its letter to the two European commissioners responsible for the two acts immediately raises the concern that imposing strict liability on tech firms “would be disproportionate and ill-suited to the properties of software”. And, of course, it could have “a chilling effect” on “innovation”.

Ah yes. That would be the same innovation that led to the Cambridge Analytica scandal and Russian online meddling in 2016’s US presidential election and UK Brexit referendum and enabled the livestreaming of mass shootings. The same innovation behind the recommendation engines that radicalised extremists and directed “10 depression pins you might like” to a troubled teenager who subsequently ended her own life.

It’s difficult to decide which of the two assertions made by the CCIA – that strict liability is “ill suited” to software or that “innovation” is the defining characteristic of the industry – is the more preposterous. For more than 50 years, the tech industry has been granted a latitude extended to no other industry, namely avoidance of legal liability for the innumerable deficiencies and vulnerabilities of its main product or the harm that those flaws cause.

What is even more remarkable, though, is how the tech companies’ claim to be the sole masters of “innovation” has been taken at its face value for so long. But now two eminent competition lawyers, Ariel Ezrachi and Maurice Stucke, have called the companies’ bluff. In a remarkable new book, How Big-Tech Barons Smash Innovation – And How to Strike Back, they explain how the only kinds of innovation tech companies tolerate is that which aligns with their own interests. They reveal how tech firms are ruthless in stifling disruptive or threatening innovations, either by pre-emptive acquisition or naked copycatting, and that their dominance of search engines and social media platforms restricts the visibility of promising innovations that might be competitively or societally useful. As an antidote to tech puffery, the book will be hard to beat. It should be required reading for everyone at Ofcom, the Competition and Markets Authority and the DCMS. And from now on “innovation for whom?” should be the first question to any tech booster lecturing you about innovation.

What I’ve been reading

The web of time
The Thorny Problem of Keeping the Internet’s Time is a fascinating New Yorker essay by Nate Hopper on the genius who, many years ago, created the arcane software system that synchronises the network’s clocks.

Trussed up
Project Fear 3.0 is a fine blogpost by Adam Tooze on criticism of the current Tory administration.

Tech’s progress
Ascension is a thoughtful essay by Drew Austin on how our relationship to digital technology has changed in the period 2019-2022.