EU AI Act Enters Enforcement: First Bans, First Fines, and US Friction
What happened
The EU AI Act moved into active enforcement in February 2026, triggering the first assessments of General Purpose AI (GPAI) models as "systemic risk" systems. Simultaneously, Washington's decision to restrict foreign nationals from accessing its most advanced AI models created a transatlantic friction that surfaced at the G7 Évian summit.
How it developed
The EU AI Act, passed in 2024, applies to AI systems used in Europe. Its most controversial provisions cover "General Purpose AI" (GPAI) models — large foundation models from US companies — which must submit to systemic risk assessments if they exceed defined compute thresholds. February 2026 marked the first enforcement deadline.
EU AI Act GPAI enforcement provisions come into force (February 2026)
EU designates several US foundation models as GPAI systems requiring systemic risk assessment
US announces restrictions blocking foreign nationals from accessing its most advanced AI models
European allies at G7 Évian express alarm at being potentially cut off from US AI (June 15–16)
AI company CEOs from Anthropic, OpenAI, Google, Mistral join G7 working lunch (June 16)
G7 issues statement on AI safety coordination — without binding commitments
- EU AI Act enforcement began in February 2026 with its first GPAI designations
- The US decision to restrict foreign national access to advanced AI models alarmed G7 allies
- AI governance coordination is now formally on the G7 agenda but without binding agreements
Unstated assumptions
- •US access restrictions serve national security goals rather than commercial protectionism
- •The EU AI Act will be enforced consistently across US and European AI companies
Whose voice is missing
- •Civil society and labour perspectives on AI deployment risks are largely absent from G7-level AI governance discussions
- •The Global South has no seat at the G7 AI governance table despite being a major AI deployment market
By outlet
"Trump administration releases preliminary agreement with Iran — AI dominates G7 sidelines"
- Framing
- AI access restrictions as an unexpected G7 fault line
- Emphasis
- European alarm at being cut off from US AI
- Downplays / omits
- Technical details of what the access restrictions cover
"Does the G7 still shape the global economy?"
- Framing
- G7 AI governance as a legitimacy problem for a declining forum
- Emphasis
- G7's inability to include the Global South in AI norm-setting
- Downplays / omits
- EU AI Act's specific enforcement mechanism
"EU AI Act: first enforcement actions and what they mean for tech firms"
- Framing
- Regulatory compliance timeline and industry reactions
- Emphasis
- GPAI designation process and systemic risk thresholds
- Downplays / omits
- Small and medium tech firm compliance burden
"EU AI Act vs US AI access rules: Europe caught between its own regulation and Washington"
- Framing
- European strategic dilemma in AI governance
- Emphasis
- Tension between EU regulation and US market dependence
- Downplays / omits
- Chinese AI development as a third option