ConductAtlas › Regulations ›

EU AI Act - High Risk Provisions

international — EU
Effective: August 2, 2026 56 platforms tracked 589 provisions indexed Enforced by: EU AI Office, National Market Surveillance Authorities

Overview

Obligations for high-risk AI systems including conformity assessments, risk management, data governance, transparency, human oversight, and EU database registration.

Penalties

Up to EUR 35 million or 7% of global annual turnover for prohibited AI practices. Up to EUR 15 million or 3% for other violations.

Key Articles & Sections

Platforms We Track Subject to

AI21 Labs
12 relevant provisions
Anthropic
63 relevant provisions
Anyscale
12 relevant provisions
Apple Intelligence
0 relevant provisions
Baseten
0 relevant provisions
Cerebras
11 relevant provisions
Cohere
21 relevant provisions
Copy.ai
6 relevant provisions
Cursor
19 relevant provisions
DeepSeek
2 relevant provisions
Descript
3 relevant provisions
ElevenLabs
4 relevant provisions
Fireworks AI
6 relevant provisions
Glean
7 relevant provisions
Google AI Studio
3 relevant provisions
Google Gemini
22 relevant provisions
Groq
13 relevant provisions
Harvey AI
0 relevant provisions
Hugging Face
18 relevant provisions
Ideogram
9 relevant provisions
Inflection AI
4 relevant provisions
Jasper AI
6 relevant provisions
Kling AI
4 relevant provisions
LangChain
0 relevant provisions
Leonardo AI
9 relevant provisions
LlamaIndex
0 relevant provisions
Luma AI
15 relevant provisions
Midjourney
23 relevant provisions
Mistral AI
29 relevant provisions
NotebookLM
0 relevant provisions
Showing 30 of 56 platforms. View all platforms →

Recent Changes Related to

ⓘ

ConductAtlas maps governance language to potentially relevant regulatory frameworks. Regulatory applicability and enforceability may vary by jurisdiction, enforcement context, and individual circumstances. This page is informational and does not constitute legal advice. Methodology

Provisions Governed by (589 across 56 platforms)

Input and Output License for Model Improvement AI21 Labs
Medium
Disclaimer of Warranties AI21 Labs
Medium
Data Subject Rights (GDPR and CCPA) AI21 Labs
Medium
Data Subject Rights (Access, Deletion, Portability) AI21 Labs
Medium
Prompt and Interaction Data Collection and Model Training Use AI21 Labs
Medium
User Indemnification Obligation AI21 Labs
Medium
Enterprise API Data Processing Agreement AI21 Labs
Medium
Agentic Use and Autonomous Action Guidelines Anthropic
Medium
Controller vs. Processor Boundary Anthropic
Medium
Model Training Opt-Out with Safety-Flagging Carve-Out Anthropic
Medium
Controller vs. Processor Distinction Anthropic
Medium
Data Controller versus Processor Scope Distinction Anthropic
Medium
Model Training Opt-Out with Safety Review Carve-Out Anthropic
Medium
Output Rights Assignment to Users Anthropic
Medium
Age Restriction and Parental Consent Requirement Anthropic
Medium
Output Ownership and No Model Training Anthropic
Medium
Controller vs. Processor Split for Enterprise Users Anthropic
Medium
Government Customer Use-Restriction Carve-Out Anthropic
Medium
Controller vs Processor Scope Limitation Anthropic
Medium
User Rights Framework (Access, Correction, Objection, Portability) Anthropic
Medium
Liability Cap at $100 or Fees Paid Anthropic
Medium
Regional Privacy Supplements (GDPR, UK, California, Canada, Brazil, South Korea) Anthropic
Medium
Intellectual Property Assignment of Outputs Anthropic
Medium
Limitation of Liability and Disclaimer of Warranties Anthropic
Medium
Prohibition on Securities and Financial Advice Reliance Anthropic
Medium
Safeguards Team Monitoring and Enforcement Anthropic
Medium
Agentic AI and Model Context Protocol (MCP) Additional Guidelines Anthropic
Medium
Limitation of Liability Anthropic
Medium
Limitation of Liability and Warranty Disclaimer Anthropic
Medium
Model Training Opt-Out with Feedback and Safety Exceptions Anthropic
Medium

Showing 30 of 589 provisions. View all →

Official Source

View official regulation text →

Get alerted when platforms change their policies — including -relevant provisions.

Subscribe to Watcher — $9.99/mo

Frequently Asked Questions

What does require?

Which platforms does apply to?

ConductAtlas tracks -relevant provisions across 56 platforms. Each platform's specific provisions are classified by severity and mapped to requirements.

How does ConductAtlas monitor compliance?

ConductAtlas captures policy documents daily, classifies provisions by regulatory framework, and flags changes that affect obligations. Every change is archived with cryptographic verification.