On 2 August 2026 the EU AI Act comes fully into force. Four risk levels, banned practices, fines up to 7% of a company's global turnover.
- On 2 August 2026 the EU AI Act becomes fully applicable: a risk-based framework with 4 levels (published before the date).
- 8 banned practices (incl. social scoring), strict obligations for high risk, disclosure for chatbots.
- Fines reach up to 7% of global turnover; high-risk embedded systems have deadlines through 2027-2028.
The EU AI Act - the first comprehensive AI law in a major jurisdiction - is entering its key phase. On 2 August 2026 the framework becomes fully applicable (I'm writing this before the date - the event is still ahead). The logic is simple. The bigger the risk, the stricter the rules.
The effect is global, not just European. A company that wants access to Europe - Anthropic, OpenAI, Google - has to follow these rules. And once it builds them in for Europe, it builds them in for everyone. The EU exports its standard - and it reaches you too, even outside Europe.
On paper the framework is sensible - it grades oversight by harm instead of banning everything across the board. But a good framework is only worth as much as its enforcement. Eight banned practices and a fine up to 7% mean something only if someone actually goes after them. Until then it's good intentions with a deadline.