EU Delays Tesla FSD Vote to December Over Belgian Speed Data

The European Union's Technical Committee on Motor Vehicles will not vote on approving Tesla's Full Self-Driving (Supervised) system across the bloc on Oct. 6 as Tesla told customers it would. The panel's next session is in December, and only seven of the EU's 27 member states recognize the system, according to Electrek and Teslarati, which independently reviewed the committee's Sept. 11 draft agenda.

The delay lands one week after a Brussels pedestrian-safety group tested the software on the city's own streets, and three weeks after Tesla filed its own safety data with EU regulators. Both documents bear on the same question the December vote will decide: whether Tesla's evidence, not its marketing, clears a bloc-wide bar of 15 states representing 65 percent of the EU's population, roughly 292 million people.

What the Netherlands already approved, and who followed

The Dutch vehicle authority RDW granted FSD (Supervised) the EU's first type approval on April 10, using Article 39 of Regulation 2018/858, an exemption path for technology that doesn't fit existing vehicle rules. Six more states have since recognized that approval: Denmark, Belgium, Estonia, Lithuania, Slovenia and Czechia, Teslarati reported. Together the seven hold about 53 million people, 12 percent of the bloc.

Sweden wants Tesla to remove the software's ability to drive above the posted speed limit before it will sign on. France has rejected the Dutch approval outright, citing speeding and driver attention in cities. Germany, Italy and Spain, which together hold 58 percent of the EU's population, have not recognized it either. A qualified-majority vote needs 15 states and 65 percent of the population, so December's outcome turns on some combination of those three.

Tesla's own numbers moved three times in three months

Tesla published a safety dashboard on Sept. 1 claiming FSD (Supervised) is 4.1 times less likely to be involved in a collision than manual driving, based on more than 100 million kilometers driven across the five countries then approved. The underlying document reports three highway collisions and nine non-highway collisions for FSD-equipped cars, against 137 and 490 for manually driven Teslas over the same distance, Automotive World's analysis found.

The multiplier itself has not held still. In June, on Netherlands-only data, Tesla cited a 3.5x improvement. By late July, across 65 million kilometers, that had risen to 5.2x. September's 100-million-kilometer figure came back down to 4.1x. Tesla's own methodology note says the comparison does not control for traffic density, weather, road surface, time of day or scenario complexity.

More telling than the swing is the paper trail. The document Tesla filed Sept. 1 has every dated measurement window closing in 2025: mileage tables run through calendar 2025, emergency-braking figures stop Oct. 31, 2025, and lane-change metrics end in early December 2025. The most recent dated entry in the entire filing is Dec. 5, 2025, nine months before the document was submitted and outside the April-to-August 2026 window the 4.1x figure is supposed to cover. Tesla has not said why a filing meant to support current performance contains no 2026 measurements.

A Brussels group tested it on the streets Tesla's filing doesn't cover

Johanna.be, a Brussels nonprofit focused on pedestrian and cyclist safety, spent three days in July and about 400 kilometers testing FSD (Supervised) on the city's low-speed streets, then filed its findings with the Dutch RDW and Flemish ministries on Sept. 7. The group's own report, which I read directly, records the car exceeding the limit in 55 percent of tested 30 km/h zones, averaging 44 km/h where the law caps it at 30. The World Health Organization's figures, cited in the report, put a pedestrian's fatality risk at a 44 km/h impact at nearly six times the risk at 30 km/h.

In the 20 km/h woonerven, Belgium's shared residential streets, FSD exceeded the limit in 42 percent of tested locations. More striking than the speed itself is what the dashboard showed drivers: in a majority of 30 km/h zones, the car's cameras correctly read a 30 km/h sign, then displayed 50 km/h as the applicable limit moments later. Johanna.be's testers also recorded FSD attempting to overtake cyclists on fietsstraten, streets where Belgian law bans motor vehicles from passing bikes at all, a rule the group says the system appears to conflate with the different right-of-way norms of neighboring countries. The same report credits FSD with yielding proactively to pedestrians and cyclists in the majority of encounters it logged.

What would change the outcome in December

None of this proves FSD is unsafe in aggregate; Tesla's non-highway collision count, even taken at face value, is a fraction of the manual-driving figure over the same distance. What it shows is that the two documents in front of European regulators don't overlap. Tesla's number describes a highway-heavy sample with no 2026 data; Johanna.be's describes exactly the low-speed, pedestrian-dense streets Tesla's filing is weakest on. A December vote that weighs both would need Tesla to publish a 2026-dated update, or a fourth country to break the standoff among Germany, Italy and Spain. Neither has happened yet, and Musk's own reaction to the delay, posted as a single word, "Sigh," according to Teslarati, suggests Tesla did not expect to need either.

The pattern echoes the NHTSA audit into Tesla's Cybercab certification, where regulators are also testing a self-certification against a paper trail rather than a demonstration. It's a different kind of scrutiny than the one facing Waymo's slower, operator-supervised rollout in Singapore: Tesla is asking to be trusted on evidence it filed itself, and December is when Europe decides whether that evidence, on its own dates, still supports the claim.

1. June (Netherlands only)3.5 x safer than manual driving
2. Late July (65M km)5.2 x safer than manual driving
3. September (100M km)4.1 x safer than manual driving

Source: Automotive World analysis of Tesla's Article 39 safety filings, accessed 2026-09-25

Country	Position	Share of EU population
Netherlands, Denmark, Belgium, Estonia, Lithuania, Slovenia, Czechia	Recognize the RDW approval	12%
Sweden	Opposed unless speed-limit override is removed	-
France	Rejected the approval	-
Germany, Italy, Spain	Undecided	58%