The EU AI Act reaches full enforcement on August 2, 2026, activating the highest-risk provisions and penalty structures that apply to foundation models like Anthropic's Claude. With just 15 days remaining, fewer than 40% of in-scope companies have completed mandatory conformity assessments, leaving significant compliance gaps across the industry.

What Changes When the EU AI Act Goes Fully Live?

The EU AI Act has rolled out in phases since February 2025, but August 2 marks the moment when every provision becomes enforceable. This final phase activates three critical obligations that directly affect companies deploying or providing foundation models like Claude, GPT, and Gemini.

• General-Purpose AI Model Obligations: Any provider of a foundation model must now comply with transparency requirements, documentation standards, and copyright policy disclosures. Anthropic and other model providers must demonstrate they've met these obligations, and companies deploying Claude need to verify their provider's compliance documentation.
• High-Risk AI Conformity Assessments: Systems deployed before May 2025 that are still in operation must now have a conformity assessment on file. This applies to AI used in critical infrastructure, education, employment, essential services, law enforcement, migration, and democratic processes.
• Full Penalty Structure Activation: Violations now carry fines of up to €35 million or 7% of global annual turnover for the most serious breaches, including violations of prohibited practices and GPAI obligations. Lower-tier violations carry fines of €15 million or 3%.

The territorial scope follows the same logic as GDPR: if your AI system processes data from people in the EU or if its output is used in the EU, you're covered, regardless of where your company is physically located.

Why Are So Many Companies Still Unprepared?

A June 2026 study found that fewer than 40% of companies with in-scope AI systems had completed mandatory conformity assessments, and under 30% had documentation ready for GPAI obligations. The compliance gaps are concentrated in specific areas that many organizations underestimated.

The most common missing pieces include technical documentation for training data sources, risk management system documentation, and human oversight protocols. Many enterprise teams assumed that using a compliant model provider like Anthropic would handle all their obligations, but the EU AI Act places requirements on deployers as well as model providers.

Adding to the pressure, companies are now operating under what experts call "regulatory simultaneity." China's agent regulations became enforceable on July 15, Illinois mandates third-party safety audits, and the UN Global Dialogue on AI Governance is developing international standards. No single jurisdiction's requirements serve as a proxy for the others, forcing compliance teams to navigate multiple, overlapping frameworks simultaneously.

How to Prepare Your Organization for August 2 Compliance

• Build an AI System Inventory: Identify every AI system in your organization that touches EU users or EU data. If you don't have a complete inventory, building one now is the first critical step. The August 2 deadline doesn't require perfection, but it does require demonstrable effort toward compliance.
• Verify Model Provider Documentation: Request EU AI Act compliance documentation from your foundation model providers. Major providers like Anthropic, OpenAI, and Google have published compliance documentation. Verify what's covered and what additional obligations fall on your organization as a deployer.
• Prioritize High-Impact Systems: For high-risk systems, prioritize conformity assessments for systems that affect the most people or handle the most sensitive data. EU regulators have signaled they'll prioritize enforcement against high-impact violations rather than technical documentation gaps.
• Document Your Progress: The Act is enforced through transparency. If you can show regulators what you've done and what you're working on, you're in a much better position than if you can't demonstrate any effort. Incomplete documentation is better than none.

For companies using Claude or other foundation models in EU-facing applications, the stakes are particularly high. Anthropic has published compliance documentation for the EU AI Act, but deploying organizations must still conduct their own risk assessments and maintain documentation showing how they're meeting the Act's transparency and oversight requirements.

The August 2 deadline is no longer theoretical. With 15 days remaining, companies that have delayed compliance face a compressed timeline to demonstrate effort, even if they can't achieve perfect compliance by the deadline. The regulators' focus on high-impact violations suggests that organizations showing good-faith compliance efforts will face less scrutiny than those showing no effort at all.