Brussels Study Finds Tesla FSD Sped on 55% of Tested Road Segments

Key Facts

Belgian road-safety group Johanna.be found that Tesla's driver-supervised FSD exceeded the 30 km/h limit on 16 of 29 road segments it tested in Brussels, or 55% of the sample. Speed averaged 44 km/h on segments where the limit was exceeded during testing conducted in July 2026. The findings come as a vote on extending permission to use the system across the EU could take place on 2026-10-06, Reuters reported. Dutch vehicle authority RDW has not announced a change to its approval for use in the Netherlands. The test raises a specific question for regulators: does the system recognize the legal limit, show it to the driver and then obey it when selecting a speed? The driver remains responsible for watching the road and intervening while FSD is in use.

Johanna.be used one Tesla Model 3 running a European version of FSD on selected roads subject to a 30 km/h limit. The car passed through the roads 59 times, but the 55% finding was calculated across 29 distinct segments rather than individual runs. The group set the driver's speed offset to 0% and disabled the feature that adjusts maximum speed according to traffic context. It recorded the legal limit, the limit the car recognized, the limit shown to the driver and the speed the system selected on each segment. Repeat passes on some roads helped the group examine behaviour, but did not turn its sample into a measure of performance on all European roads. The result describes this car and software version under the selected street and test conditions.

The vehicle's display shows why recognizing a sign and controlling speed need to be examined separately. In cases the group documented, FSD recognized a 30 km/h sign, then showed the driver an applicable limit of 50 km/h and sometimes selected a speed above the legal limit. The study says the displayed limit was wrong on about 90% of the tested Brussels segments. A driver relying on that figure could be slower to notice the need to intervene, even though the duty to supervise driving remains with that person. Sign recognition alone is insufficient if the speed the system chooses does not follow the limit it recognized. Reviewing this behaviour requires checking the observed sign, displayed limit and actual speed on the same segment.

The sample included 21 Brussels segments with clear signs marking a 30 km/h zone. Of 17 such segments tested in traffic that let the car select its speed, the group recorded speeding on 10, or 59%, averaging 46 km/h where the limit was exceeded. In shared residential areas with a 20 km/h limit, it found speeding in 5 of 12 areas, or 42%. The test also documented FSD attempting to pass cyclists on streets where Belgian rules prohibit cars from doing so. That breakdown shows the group's observations were not confined to one type of low-speed street. The samples and road conditions differ, however, so those percentages alone cannot describe performance on every trip.

RDW approved driver-supervised FSD for use in the Netherlands in April 2026, with possible later recognition in other EU countries. The authority describes it as advanced driver assistance, not a car that relieves its driver of monitoring the road and being ready to intervene. RDW said its assessment lasted more than 18 months and covered public roads and test tracks. In a later account, it cited more than 3000 testing hours, more than 1000 runs and European driving data covering 1.8 million kilometres. That broad programme assessed the system in varied conditions, while Johanna.be focused on specific rules on low-speed streets. The practical question is how fully the broader testing addressed the segments where the Belgian group observed an incorrect limit or speed.

Tesla published a different analysis of speed behaviour to support its European regulatory application for FSD. It selected 708 European driving samples under conditions that included other vehicles travelling in the same direction and a maximum speed set above the limit identified by the system. After excluding low-speed cases, Tesla said the car travelled at or below the median speed of surrounding traffic in 98% of 680 samples. That comparison measures the car against traffic flow; the Brussels test measures compliance with the legal limit on particular streets. Surrounding vehicles may travel faster than the legal limit, so matching their pace cannot settle whether that limit was obeyed. The different sampling rules also make checks of low-speed street signs relevant to understanding the gap between the findings.

RDW told Reuters it had received Johanna.be's report and would consider it through its monitoring and surveillance procedures. Reuters also reported a Flemish transport-ministry spokesperson's reminder that the driver must intervene if the system makes an error or applies a traffic rule incorrectly. Johanna.be recorded a positive aspect of its test, describing FSD as cautious around pedestrians and cyclists in observed encounters. Yielding to a vulnerable road user, however, does not establish whether the car obeyed the legal speed on another segment. A driver can correct the selected speed during a trip, but reliance on that intervention makes the clarity of the displayed limit consequential. Any decision on software changes or operating conditions rests with regulators assessing the observations and subsequent testing.

For a TSLA holder, the financial issue is the number of countries where FSD could be used if approval expands within the EU. Shares traded at $380.12 on 2026-09-23 and $377.94 on 2026-09-24, according to EL7 data. Comparing those two closes cannot isolate any effect of the Brussels test from other influences on Tesla trading. A favourable vote could widen permitted use under the approval route RDW has described. Continued compliance questions, conversely, could affect the timing or conditions of an expansion. That is a conditional question about regulatory access; the group offered no estimate of a change to Tesla's sales or earnings from its test.

The next regulatory milestone may come if the application to expand FSD approval across the EU reaches a vote on 2026-10-06, a possible date reported by Reuters. RDW says the application goes to member states and that a favourable vote would extend the system's validity across the bloc. Before then, RDW's handling of Johanna.be's findings and any explanation Tesla provides about speed limits would help indicate the review's direction. Repeat tests on the disputed segments could show whether the displayed limit and selected speed match the local rule. If a higher limit still appears after the car recognizes a lower sign, the compliance question would remain even if the system performs well in other settings. Consistent agreement among the recognized sign, display and speed would instead support a finding that the specific fault identified by the study had been addressed.