EU AI Act for Automotive: Vehicle Safety AI, ADAS & the Product-Safety Route
Carmakers embed AI in driver-assistance, perception, autonomous-driving and in-cabin systems. Under the EU AI Act (Regulation (EU) 2024/1689), AI that performs a vehicle safety function is high-risk through the product-safety route, not the Annex III list — and that route carries the Act's latest deadline. Non-safety features like in-car voice assistants sit in the lighter transparency tier.
Is it in scope?
The relevant trigger is Art. 6(1): AI is high-risk if it is a safety component of a product covered by EU harmonisation legislation in Annex I and subject to third-party conformity assessment. Vehicle type-approval and machinery legislation fall in that group, so AI in ADAS, perception and autonomous-driving safety functions can be high-risk, with obligations applying from 2 August 2027. In-cabin conversational assistants and infotainment generally attract only Art. 50 transparency, and manufacturing-line AI follows the manufacturing rules.
Typical AI use cases
- Advanced driver-assistance systems (ADAS)
- Perception and object-detection for autonomous driving
- Automated emergency braking and safety functions
- In-cabin voice assistants and infotainment
- Driver-monitoring systems
Risk classification
AI that is a safety component of a vehicle or its systems is high-risk under Art. 6(1) via the product-safety route, with the manufacturer as provider carrying Art. 9–15 — designed to be integrated into the existing type-approval and conformity framework, effective 2 August 2027. Non-safety features such as in-car assistants are typically limited-risk under Art. 50. Note that driver-monitoring that infers emotional state raises Art. 5 emotion-recognition concerns and must be scoped carefully, even though safety-driven attention monitoring can be legitimate.
Obligations to prepare for
Your exact duties also depend on whether you build or use the AI. See obligations by operator role — provider, deployer, importer, distributor or GPAI provider.
FAQ
Is ADAS or autonomous-driving AI high-risk?
Yes, where it is a safety component. Under Art. 6(1) AI that performs a safety function in a vehicle covered by EU harmonisation legislation requiring third-party conformity assessment is high-risk, with obligations applying from 2 August 2027.
Why is the deadline 2027 and not 2026?
AI regulated through the product-safety route (Art. 6(1) and Annex I) has a later start date, 2 August 2027, than the 2 August 2026 date that applies to most Annex III high-risk uses. This gives product manufacturers additional time to align conformity assessments.
Does our in-car voice assistant need to comply?
Its main duty is Art. 50 transparency — informing the occupant they are interacting with an AI system, from 2 August 2026. It is not high-risk unless it performs a vehicle safety function, though driver-monitoring designs should be checked against the Art. 5 emotion-recognition limits.
Run the free 2-minute risk checker — no sign-up.