The EU wants to set the global standard for responsible AI regulation. But it fears that complicated digital rules could block its nascent AI industry from succeeding in the intensifying race for AI dominance.
Nowhere have these rivalling ambitions clashed more than over the EU’s new general-purpose AI Code of Practice, which details the AI Act’s vague requirements for ChatGPT and the likes.
1,000 lobbyists with irreconcilable views have fought it out over 12 heated months. While companies demanded minimal requirements, the Commission promised an innovation-friendly implementation. For their part, civil society and MEPs panicked that the Code (which grants presumed compliance) would undermine an essential part of the AI Act.
The make-or-break moment has now arrived, as companies decide whether or not to sign on to the final Code.
The world’s wealthiest companies will have few qualms in signing a weak Code, which will grant them a cheap and easy path to making their most powerful technology AI Act compliant. But a more exacting Code runs the risk of having no signatories at all – making a mockery of the Commission’s business-friendly pitch.
A sigh of relief in the Berlaymont then, when ChatGPT-maker OpenAI and French challenger Mistral said they would sign last week.
Meta, by contrast, will not sign the Code.
This is good news: The tech giant’s refusal signals that the intense industry pressure failed to fully warp the Code to Meta’s liking.
Mark Zuckerberg is hell-bent on winning the AI race. He has reportedly offered salary packages of €260 million – worth more than 99.996% of European companies – to at least 10 top AI researchers to poach talent for a new “superintelligence” team.
At the same time, he and Meta have repeatedly demonstrated that they are bad-faith actors who must be dragged kicking and screaming to comply with EU laws.
In June last year, Meta halted its AI rollout in Europe after its opt-out scheme to train AI on social media posts was peppered with data protection complaints. It warned that EU rules would create a “gap in the technologies that are available in Europe,” versus the rest of the world. In April, it relaunched AI features with a minimally adjusted opt-out scheme to train on users’ posts.
The pattern is clear: First, Meta rolls out a model it knows will challenge EU laws, preferring to lawyer up rather than practise due diligence. Then it loudly accuses the EU of strangling innovation and causing legal uncertainty, threatening to pull its products from the market. When push comes to shove, it makes marginal adjustments that stretch the law to its limits.
It’s not just AI training – the company battles with all the EU’s digital laws. Meta has been clashing with the Commission over its “pay or consent” advertising model, for which it was fined €200 million in April and could still face further fines. Meta is under investigation for breaching the EU’s digital platforms law, and is separately accused of partnering with sanctioned Russian publishers.
Zuckerberg, who alone holds majority voting shares in the €1.5 trillion company, is spearheading Meta’s cynicism. After Donald Trump took office, Zuckerberg turned on a dime to dismiss fact-checking as censure and praised growing “masculine energy” in the private sector. Meta is calling for Trump to intervene against EU enforcement of digital rules, which Zuckerberg has likened to tariffs.
The list goes on. Time and again, Meta has shown it has no interest in honest and constructive dialogue. Zuckerberg is moving fast to break things, including the EU’s AI rulebook. Meta’s condemnation should be a relief.
“What will, however, not be voluntary for Meta is to be compliant with the AI Act by 2 August,” said a Commission spokesperson today.
Roundup
Tokyo teamup – Strategic joint procurement of critical raw materials and tighter business-to-business links are set to form the backbone of a new ‘EU-Japan Competitiveness Alliance,’ according to a draft summit communiqué seen by Euractiv, which is expected to be launched on Wednesday as EU leaders visit Asia.
Fungicide-icide – An EU expert group has confirmed it has decided against the use of the fungicide potassium phosphonate in organic agriculture because the chemical is synthetic and leaves traces in food that persist for years.
EU speaks on Israel aid killings – EU foreign policy chief Kaja Kallas told her Israeli counterpart Tuesday that Israel’s military “must stop” killing civilians at aid distribution points in Gaza, after the World Health Organization said its facilities in Gaza had been targeted by Israeli forces.
Across Europe
Black gold in the Baltic – The Polish arm of Canadian-founded Central European Petroleum has announced Poland’s largest-ever oil and gas discovery near Świnoujście off the Baltic coast, which could hold over 33 million tons of market-grade gas and provide Poland and the EU with much-needed gas.
Anti-corruption changes – Ukraine MPs approved amendments to remove the independence of two Ukrainian anti-corruption bodies Tuesday, a day after the arrest of one of the agency’s officials. The bodies will be placed under control of the prosecutor general, who is appointed by the president.
Iran persists with nuclear programme – Iran is scheduled to meet Britain, France and Germany in Istanbul on Friday, but the nation has said it has no plans to abandon its nuclear programme, including uranium enrichment, despite the “severe” damage caused by US strikes to its facilities. Iran will also meet with Chinese and Russian representatives on Tuesday.