China Moves to Seize Autonomous Driving Data as Crash Evidence

China's Ministry of Public Security has proposed sweeping amendments to its traffic accident handling procedures that would explicitly authorize police to extract data from autonomous driving data recording systems (DSSAD), vehicle dashcams, and other electronic sources as official evidence in crash investigations. The draft regulation, published on May 8 for public comment, signals Beijing's push to build a legal framework around the rapidly expanding autonomous vehicle industry.

The proposed changes to the Road Traffic Accident Handling Procedures come at a critical moment, as Chinese automakers including BYD, NIO, XPeng, and Huawei-backed AITO aggressively roll out advanced driver-assistance systems (ADAS) and semi-autonomous features across millions of vehicles on Chinese roads.

Key Takeaways From the Proposed Amendments

• Police can now formally collect data from autonomous driving data recording systems and vehicle event data recorders (EDRs) as admissible electronic evidence
• Remote accident processing via video calls becomes officially permitted, allowing a single traffic officer to handle cases online
• 2 new evidence-gathering methods added: video recording and 3D scene reconstruction for crash sites
• Digital delivery of accident determination documents replaces mandatory in-person signatures
• Electronic verification of vehicle inspection and insurance status replaces physical sticker checks
• The regulation covers 114 total articles, with 9 articles being modified in this revision

Why Autonomous Driving Data Access Matters

The most significant change for the global autonomous vehicle industry is the explicit legal basis for police to access self-driving system logs. Unlike the United States, where the National Highway Traffic Safety Administration (NHTSA) has gradually built its authority to demand AV data through a patchwork of investigations and recalls — most notably involving Tesla's Autopilot crashes — China is proactively codifying this authority into its procedural rules.

This matters because determining fault in accidents involving autonomous or semi-autonomous vehicles requires understanding what the AI system 'saw,' what decisions it made, and whether the human driver was properly engaged. Traditional accident investigation methods — skid marks, witness testimony, traffic camera footage — are often insufficient for crashes where software is a potential factor.

The regulation positions China's law enforcement to demand granular telemetry data, including sensor inputs, algorithmic decision logs, and control handoff timestamps. For automakers operating in China, this means their black box equivalent systems must be designed for regulatory accessibility from the ground up.

How China's Approach Compares to Western Regulations

The United States and European Union have taken different paths toward regulating AV crash data. In the U.S., NHTSA's Standing General Order (SGO) 2021-01 requires manufacturers and operators of vehicles equipped with Level 2 ADAS and higher to report crashes, but the data-sharing framework remains largely voluntary and investigation-driven.

The EU's General Safety Regulation, which took effect in July 2024, mandates event data recorders in all new vehicles. However, access to that data for law enforcement purposes varies by member state and often requires court orders.

China's proposed approach is notably more direct:

• Scope: Covers all vehicles with driving recorders or autonomous systems, not just those above a certain automation level
• Authority: Grants frontline traffic police — not just federal investigators — the power to collect electronic data
• Speed: Data can be collected at the scene or shortly after, without requiring separate judicial authorization
• Integration: Electronic evidence is treated on par with physical evidence like photographs and witness statements

This regulatory directness reflects China's centralized governance model and its strategic priority of leading the global autonomous driving market, currently valued at over $75 billion and projected to exceed $400 billion by 2035 according to McKinsey estimates.

Modernizing Crash Investigations With 3D Scene Reconstruction

Beyond data access, the proposed amendments introduce real-world 3D scene capture as an officially recognized evidence-gathering technique. This aligns with technologies already deployed by companies like Matterport and FARO Technologies in Western forensic contexts, but China is embedding the method directly into its national traffic procedures.

The addition of 3D reconstruction alongside traditional photography and sketching gives investigators a volumetric record of crash scenes. This is particularly valuable for autonomous vehicle incidents, where the precise positioning of vehicles, pedestrians, and infrastructure — measured in centimeters — can determine whether an AI system responded appropriately.

The regulation also acknowledges practical realities by allowing investigators to clear accident scenes first and then create supplementary diagrams and investigation notes afterward. This reduces traffic disruption while preserving evidentiary integrity through digital capture.

Remote Processing Signals Digital-First Enforcement

Another notable modernization is the formalization of remote accident processing. Under the proposed rules, minor property-damage accidents can be handled by a single traffic officer via internet or video connection, rather than requiring physical presence at the scene.

This change has several implications:

• Efficiency: Reduces response times and officer workload in a country with over 340 million registered vehicles
• Digital signatures eliminated: Accident determination documents can be delivered electronically, removing the need for in-person signing
• AI-assisted triage: The framework opens the door for future integration of AI-powered preliminary accident assessment tools
• Scalability: Enables handling of the growing volume of minor ADAS-related incidents without proportional staffing increases

For context, China reported approximately 250,000 traffic fatalities in recent years according to World Health Organization data, making efficient accident processing a public safety priority.

What This Means for Automakers and Tech Companies

Foreign and domestic automakers selling vehicles with ADAS or autonomous capabilities in China should prepare for increased data transparency requirements. Tesla, which already faced scrutiny in China over brake-failure allegations in 2021 — a case that partly drove the company to open a Shanghai data center — may find the new rules formalize what was already an informal expectation.

Chinese EV makers are likely better positioned to comply. BYD's DiPilot, XPeng's XNGP, and Huawei's ADS 2.0 systems already operate under China's data localization requirements. The new regulation simply adds a clearer legal mechanism for police to access that locally stored data.

For Western autonomous driving companies eyeing the Chinese market — including Waymo, Mobileye, and various Tier 1 suppliers — the message is clear: build your data recording and access architecture to meet Chinese law enforcement standards, or risk being locked out of the world's largest automotive market.

Looking Ahead: A Blueprint for Global AV Regulation?

China's proposed amendments could influence how other countries approach autonomous driving evidence collection. As the United Nations Economic Commission for Europe (UNECE) continues developing its WP.29 framework for automated vehicle regulations, China's practical experience with these rules will provide a real-world case study.

The public comment period for the draft regulation is now open, with final rules expected to take effect later in 2025. If adopted as proposed, China will have one of the most comprehensive and enforceable frameworks for accessing autonomous driving data during crash investigations anywhere in the world.

The broader trend is unmistakable: as AI takes a larger role behind the wheel, governments worldwide are demanding the right to look under the hood — not at the engine, but at the algorithm. China is simply moving faster than most to codify that authority into law.