Tesla made public the safety evidence it gave European regulators for its Full Self-Driving approval, a document in which every dated measurement window closes in 2025 — months before the European rollout the file is now being used to justify.
The company posted late Monday, saying it had decided to open-source one of the key pieces of evidence used to support the Netherlands’ approval ahead of a potential EU-wide vote on October 6.
Tesla states in the same post that the dashboard was shared with all EU member states in April, shortly after the Dutch approval, and that all data presented predates that month.
The document is headed “Article 39 – FSD (Supervised) Dashboard” and supports exemptions from UN Regulation 171, the standard governing Driver Control Assistance Systems.
Tesla seeks five: system-initiated manoeuvres with hands-off requests withheld on highways and on other roads, no inhibition of those manoeuvres based on detected driver state, lateral acceleration above the regulatory ceiling, and permission to exceed detected speed limits to match surrounding traffic.
Data
The document narrows that further where it states its windows. Mileage distribution covers calendar 2025.
Acceleration, horn and blinker metrics run from June 2 to September 9, 2025. Automatic emergency braking covers January 1 to October 31, 2025. Lane-change safety runs October 6 to November 26, 2025, and lane-change driver attention October 6 to December 5, 2025.
The fleet collision study is headed as quarterly data from the first quarter of 2024 to the fourth of 2025.
The latest dated observation among those is December 5, 2025.
Three sections — engineering fleet testing, fixed-route testing and the speed analysis — state no window in the document’s contents, and their tables did not render in the version retrieved for this report, so whether they contain 2026 data is not established.
756,000 Miles
The exemptions Tesla is seeking apply in Europe.
Summing the dashboard’s own European table for 2025 gives 423,183 highway miles, 116,499 arterial, 174,624 urban collector and 41,616 local access — 755,922 miles, or about 1.2 million km. A later section claims over 793,000 miles across eight countries.
The two do not add up to the same thing, though different inclusion rules — which countries, engineering versus validation driving, different freeze dates — could account for the gap.
European manually driven Teslas with active safety features contributed 8.91 billion miles over the same period. The FSD sample is 0.008% of the group it is compared against.
Tesla states that its European FSD data comes from engineering operators trained to use the system with as little human input as possible, and presents this as a reason the metrics better reflect the system’s true performance.
Tesla’s Numbers
The dashboard carries a methodological note conceding that the fleet-wide studies do not control for traffic density, weather and road surface, time of day, or scenario complexity, and that these could influence when drivers choose to engage the system.
It states the comparisons should be read as observational associations under stratified exposure conditions, not as causal estimates from a randomised experiment.
Tuesday’s post closes by saying FSD is already saving lives where it is approved. The filing makes no such claim and, on its own terms, does not support the sentence Tesla attached to it.
All three control types in the study are Tesla vehicles, so the reductions describe FSD against Tesla owners driving Teslas. The weakest baseline is 120 million miles from early Model S and Model X built from 2012, which Tesla argues is the most representative comparison to the average vehicle on European and North American roads.
The File That Stays Sealed
Reuters reported on June 15 that Tesla gave regulators in the Netherlands and Sweden self-published safety statistics during its approval push, including claims that FSD vehicles travel more than seven times farther between crashes and are up to ten times safer than average human drivers, alongside a slide asserting the system could have prevented 32,000 deaths and 1.9 million injuries.
In August, Reuters published emails obtained through a public-records request showing Tesla had pressed the Dutch authority RDW since late 2024 to keep documentation of the review permanently confidential.
RDW has declined to say how it tested the system, citing commercial secrecy, and its characterisation softened from safer than other driver assistance systems in April to at least as safe as by June.
The Vote
The Technical Committee on Motor Vehicles can extend the Dutch approval bloc-wide on a qualified majority of at least 15 states representing 65% of the EU population.
The five countries that have recognised the Dutch certificate — the Netherlands, Lithuania, Estonia, Denmark and Belgium — hold about 40 million people, about 8.9% of the EU total. Germany, France, Italy and Spain are all outside the group.
RDW presented its file to the committee on May 5; the June 30 agenda carried no vote, and the regulator said on July 30 it could not give a timeline.
Tesla said in November 2025 that RDW had committed to approval in February 2026. The regulator publicly denied any commitment, and approval came on April 10.
The European Transport Safety Council wrote to transport ministers on June 12 urging them not to recognise the approval until there is meaningful, published real-world safety evidence.
France’s transport minister said on July 22 the system would not be authorised in its current form.
The Newer Numbers
Tesla’s post pairs the dashboard with a claim that FSD is 4.1 times less likely to be involved in a crash than manual driving across 100 million km on EU roads. That figure is from the customer fleet after April and appears nowhere in the file.
It is also lower than Tesla’s own previous figure.
On July 28 the company said FSD had been over 5.2 times safer across 65 million km from April 10 to July 26, with zero collisions in 41.9 million highway km and six in 23.8 million non-highway km.
The Netherlands-only figure in June was 3.5 times.
Mileage grew 54% between those two releases while the multiple fell 21%.
