Now turning its attention to general-use AI models, such as ChatGPT, Gemini, or Claude, the European Union has activated the second part of the AI Act.
The European Union has a sovereign and ethical vision for artificial intelligence — one that sharply contrasts with the self-regulated American model and the state-driven Chinese approach.
This comes amid a sensitive geopolitical and economic climate, where companies are pouring resources into AI to avoid being left behind, and investors are flocking to safe havens. Gold prices and BTCUSD are reaching historic highs.
The EU aims to impose its standards on global AI players through “extraterritoriality,” much like it did with the GDPR.
Key measures include:
- Copyright control: All AI-generated content must respect and not infringe upon original works, ensured through certification and traceability.
- Data traceability: A ban on large-scale harvesting of personal data, with strict requirements for a documented chain of custody.
- Preemptive moderation: Mandatory filtering of hateful or discriminatory speech before publication.
These measures could be restrictive for industry players. Meta (Facebook, Instagram) firmly rejects the obligations, calling them a direct threat to its data-driven business model.
Microsoft and Google have taken a more conciliatory stance, although both have expressed concerns over the potential regulatory impact. Meanwhile, Elon Musk’s xIA agreed only to sign the safety accord, claiming that other aspects would stifle innovation.
Regulatory framework and timeline
Seen by some as a form of digital protectionism, the EU’s stance could reshape the balance of power between U.S. tech giants and Europe’s ambitions in the digital space — hence the EU needs to establish a clear framework.
The European AI Act, also known as the Artificial Intelligence Regulation, came into force on August 1, 2024. Its enforcement will gradually ramp up over the next three years, with full application expected in August 2027.
- Since February 2025, prohibitions on “unacceptable” uses of AI have been in effect.
- The section covering general-purpose AI took effect on August 2, 2025, alongside a voluntary Code of Practice.
With this new phase, the European Union will introduce a Code of Good Practices for developers of general-purpose AI (GPAI) models.
While voluntary, this code is highly incentivized and serves as a transitional framework toward the binding obligations of 2026-2027.
It’s based on three main pillars:
- Algorithms and their training sources
- System security
- Copyright in generated content
Sanctions are severe: noncompliant entities risk fines of up to 7% of their global revenues. The potential penalties are strong enough to push even foreign governments to encourage compliance — despite the lack of a formal legal obligation.
Growing opposition: innovation vs. regulation
Efforts to regulate artificial intelligence heighten tensions between the EU and tech firms. Several industry players have long opposed the EU’s regulatory push, preferring to set their own standards.
At the Techarena conference in Stockholm, Meta and Google’s DeepMind executives criticized European approach, arguing that it undermines the agility and experimentation needed in a fast-evolving field.
Their primary concern: Europe could become hostile to innovation, driving talent and capital to regions with looser regulations, such as the United States.
This sentiment isn’t limited to U.S. firms. European tech stakeholders — including Spotify, venture capital firms like Lightspeed (a backer of Mistral), and industrial players like SAP and Ericsson—warn that the rules are overly complex and fragmented.
They are calling for clearer and fairer regulations that truly safeguard consumers without hampering startups. Many European startups and SMEs (Small and Medium-sized Enterprises) worry about being placed at a legal and financial disadvantage compared to larger, better-resourced corporations.
Some European unicorns are even considering relocating to jurisdictions with fewer constraints if they can’t afford to comply with all E.U. and its member states directives.
The question now is whether the EU can regulate AI without sacrificing competitiveness — fostering European champions while safeguarding its digital sovereignty. The answer may emerge in the coming months and years.
