Article 104: Amendment to Regulation (EU) No 168/2013

Article 104 extends the AI Act's reach into Regulation (EU) No 168/2013, which governs the type-approval and market surveillance of two- or three-wheel vehicles and quadricycles. This covers a wide vehicle range: mopeds (L1e), motorcycles (L3e–L5e), motorised tricycles (L2e, L5e), and quadricycles (L6e, L7e) — essentially everything from e-scooters to heavy touring motorcycles to light quadricycles.

As these vehicles increasingly incorporate AI-powered features — rider assistance systems, stability management, autonomous low-speed manoeuvring, and collision avoidance — the amendment ensures that AI Act requirements are assessed as part of the existing type-approval process. This follows the Article 8 integration model: AI obligations do not run on a separate track but are woven into the product certification the manufacturer already undergoes.

Given that rider safety is uniquely critical for two-wheelers (where the physical vulnerability of riders is far higher than for enclosed vehicles), the risk management and human oversight requirements of the AI Act carry particular weight in this sector.

• Article 8 — Integration with sector legislation: the master provision requiring AI Act compliance to be assessed through existing product conformity assessment for Annex I products. Article 104 activates this for Regulation 168/2013 vehicles.
• Annex I — Union harmonisation legislation: after the amendment, two/three-wheel vehicles and quadricycles with AI safety components fall under the integrated AI Act assessment regime.
• Article 9 — Risk management: particularly important for rider assistance AI, where system failure can directly result in rider injury or death.
• Article 14 — Human oversight: riders must retain meaningful control over vehicle behaviour; AI systems must be designed so that the rider can override or disengage automated features.
• Article 113 — Entry into force and application dates: confirms when Article 104 becomes enforceable.

For vehicle OEMs: - Catalogue all AI-powered features in your product lines: adaptive cruise control, cornering ABS with AI, blind-spot detection, predictive stability systems, and any autonomous manoeuvring capabilities. - For each AI feature, prepare documentation that addresses AI Act requirements — risk management (Article 9), data governance (Article 10), technical documentation (Article 11), logging (Article 12), and human oversight (Article 14). - Engage your type-approval authority early to understand how they will assess AI components within the existing Regulation 168/2013 framework.

For component and system suppliers: - If you supply AI-based rider assistance components (e.g., radar units with ML-based detection, IMU-driven stability algorithms), document your system's AI characteristics comprehensively so OEM customers can incorporate this into their type-approval dossiers. - Provide clear performance envelopes: under what conditions the AI performs reliably, known limitations (e.g., rain, night, unusual road surfaces), and failure mode behaviour.

Key rider safety consideration: - The physical vulnerability of two-wheeler riders means AI system failures carry disproportionate safety consequences. Risk management must address motorcycle-specific scenarios: wet roads, gravel, mixed traffic, and rider input that may conflict with AI recommendations.