Steve Dempsey: Big Tech bleating about EU AI rules little more than a fear of a basic level of oversight and respect for copyright

Irish Independent

2 days ago

At the heart of much of the discussion here is whether the need for AI innovation trumps existing copyright laws.
The US sees itself in an AI race with China, while Europe has been more focused on protecting citizens and existing rights.
The European Commission recently published implementation guidelines relating to the EU AI act.
These include details of legal obligations for the safe use of AI, copyright protections for creators, and transparency rules around how AI models are trained.
As Europe has a track record of creating de facto rules for the West around tech legislation, it's worth understanding how these implementation guidelines have been greeted.
Last week a consortium that represents rights holders from across the media, music, film & TV, books and publishing and art worlds came out against the AI guidelines. Ironically, there isn't a creative name among the host of acronyms representing the creative industries. There's AEPO-ARTIS, BIEM, CISAC, ECSA, FIM, GESAC, ICMP, IMPALA and more.
Their point is clear, though. In an open letter, they claim that the European Commission's official guidance on the copyright and transparency obligations contained in the EU AI Act favours tech companies over creators and copyright owners.
Their concern is that the new AI regulations will solely benefit the AI companies that scrape their copyrighted content without permission to build and train models.
The letter says: 'We are contending with the seriously detrimental situation of generative AI companies taking our content without authorisation on an industrial scale in order to develop their AI models. Their actions result in illegal commercial gains and unfair competitive advantages for their AI models, services, and products, in violation of European copyright laws.'
Big tech, which seems to have more lobbying muscle than coding muscle these days, is not presenting a unified front. Google has said it will sign the EU's AI code of practice but warned that the Act and the Code could make Europe an AI laggard.
Kent Walker, president of global affairs and chief legal officer at Google's parent company Alphabet, ominously warned: 'Departures from EU copyright law, steps that slow approvals, or requirements that expose trade secrets could chill European model development and deployment, harming Europe's competitiveness.'
OpenAI and the French artificial intelligence company, Mistral are also onboard. And Microsoft will more than likely sign too.
But Meta, Facebook's parent company, is against the code. They believe it introduces a number of legal uncertainties for model developers and measures that go beyond the scope of the AI Act.
Like Google, they're warning that this will throttle the development and deployment of frontier AI models in Europe, and stunt European companies looking to build businesses on top of them. Facebook knows all about how to use FOMO.
And it's working. There's been another open letter, this time from the chief executives of large European companies, including Airbus and BNP Paribas, urging a two-year pause by Brussels and warning that unclear and overlapping regulations were threatening the bloc's competitiveness in the global AI race.
With all these talking heads, commercial imperative and AI hype cycle, it's easy to forget what all this hot air is about. The issue here is Article 53 of the AI act that introduces transparency into the heart of general-purpose AI model deployment.
This article stipulates that AI providers must create and maintain detailed technical documentation covering the AI model's design, development, training data, evaluation, testing, intended tasks, architecture, licensing and energy metrics. All of this must be available to the EU AI Office and national authorities on request. It also must be available in relation to any other downstream systems that integrate the model in question.
Article 53 also ensures model providers adhere to EU copyright law and must publicly publish a detailed summary of the training data used. This aims to shed light on datasets, sources, and potential inclusion of copyrighted material.
So really, all this quibbling boils down to a level of transparency, societal oversight and a respect for copyright.
It's understandable that technology companies are bristling. China isn't tying itself up in this level of bureaucracy, right? The EU's history with tech regulation, such as the GDPR, have often set up roadblocks for users rather than truly protecting privacy.
And there's a significant opportunity cost to complying with this level of oversight. How is big tech supposed to move fast and break things with European technocrats looking over their shoulders?
But then again, maybe that's the point. When it comes to a technology that might take all our jobs or wipe us all out – depending on who you talk to – maybe a bit of technocratic oversight isn't a bad thing?
We know from recent history what happens if Silicon Valley's needs are put ahead of society's. Perhaps the artists and creators who have warned against favouring big tech capital over copyright aren't just protecting their own livelihoods. They're doing us all a favour.