The EU AI Act's enforcement launched on February 2, 2025, targeting "unacceptable risk" AI systems such as social scoring and biometric surveillance. Fines can hit €35 million or 7% of global turnover. Broader regulations for high-risk AI practices will commence in August 2026, with general-purpose AI rules effective in August 2025. Many startups express readiness concerns, lacking proper risk classifications. Compliance costs may deter innovation, but clearer governance could yield long-term benefits.