Reuters found that this ratio was inflated by roughly three times due to a basic methodological error. Tesla had compared the rate of its own airbag-deployment crashes—a narrow, severe subset of incidents—against a sweeping federal dataset that includes all tow-away crashes, regardless of severity. The federal data already breaks out airbag-deployment crashes as a separate category, making Tesla's comparison statistically invalid .

The reporting went beyond a critique of the numbers. Reuters interviewed 11 independent traffic-safety researchers; 10 of them stated the statistics amounted to misleading marketing, not reliable safety data . Further, nine former Tesla data labelers and a former self-driving engineer told Reuters they would not trust the system to drive them, citing regular failures with emergency vehicles, motorcyclists, and construction zones .

Although that Reuters investigation focused on Tesla's public-facing marketing, a follow-up report on June 15 revealed through public records requests that Tesla had presented these same self-published claims—including an assertion that FSD "could have saved 32,000 lives"—directly to regulators in Sweden and the Netherlands during the lobbying process for European approval .

The RDW's 18-Month Testing Regimen

The Dutch vehicle authority, RDW, granted Tesla a provisional type approval under UN R-171 (Driver Control Assistance Systems) on April 10, 2026, making the Netherlands the first European Union country to authorize the system on public roads . The approval did not come quickly. According to the RDW, the system was "extensively examined and tested for more than one and a half years on our test track and on public roads" . Multiple reports across the automotive press cite more than 18 months of testing, roughly 3,000 hours on a closed test track, 4,500 test scenarios, and approximately 1.6 to 1.8 million kilometers driven on European roads .

The approval remains provisional and currently applies only to vehicles equipped with Tesla's Hardware 4 (AI4) computer in the Netherlands. The RDW is pursuing EU-wide authorization on Tesla's behalf, which will require a vote from member states representing 55% of the EU's population .

The Minister's Defense: Independent Verification

Facing questions in parliament from Labour (PvdA) MP Habtamu de Hoop, Minister Karremans drew a bright line between Tesla's statistical lobbying and the RDW's technical assessment. His key argument was that the disputed safety statistics "played no part" in the final decision; the agency based its approval solely on its own independently verified testing .

Karremans refused to pause or reopen the approval process, and the RDW itself has described the system as making a "positive contribution to safety" . The minister’s stance creates a clear regulatory logic: if a nation's technical agency conducts thousands of hours of its own observation and test-track validation, it renders the manufacturer's self-published claims legally irrelevant.

The European Backdrop: Approving Alone

The Netherlands' solo-first approach has drawn sharp criticism. MP de Hoop pressed the minister on why the country approved the system before any other EU member state, citing fatal crashes in the United States involving Tesla's FSD system as evidence that broader coordination was warranted . The European Transport Safety Council (ETSC) has sent a formal letter to other member states urging them to demand public answers from sources other than Tesla before recognizing the Dutch approval .

Approximately 40,000 Teslas in the Netherlands are now using the FSD (Supervised) feature, according to Tesla Europe, which reported cars using the software had driven 23.6 million kilometers by mid-June . It is important to note that the system is classified as Level 2 driver assistance under UN regulations, meaning the human driver remains legally responsible and must maintain control at all times . Several other EU countries, including Belgium, Denmark, Estonia, and Lithuania, have already recognized the Dutch provisional approval or issued their own preliminary endorsements, creating a complex regulatory patchwork that the European Commission will eventually have to reconcile .

Reuters found that this ratio was inflated by roughly three times due to a basic methodological error. Tesla had compared the rate of its own airbag-deployment crashes—a narrow, severe subset of incidents—against a sweeping federal dataset that includes all tow-away crashes, regardless of severity. The federal data already breaks out airbag-deployment crashes as a separate category, making Tesla's comparison statistically invalid .

The reporting went beyond a critique of the numbers. Reuters interviewed 11 independent traffic-safety researchers; 10 of them stated the statistics amounted to misleading marketing, not reliable safety data . Further, nine former Tesla data labelers and a former self-driving engineer told Reuters they would not trust the system to drive them, citing regular failures with emergency vehicles, motorcyclists, and construction zones .

Although that Reuters investigation focused on Tesla's public-facing marketing, a follow-up report on June 15 revealed through public records requests that Tesla had presented these same self-published claims—including an assertion that FSD "could have saved 32,000 lives"—directly to regulators in Sweden and the Netherlands during the lobbying process for European approval .

The RDW's 18-Month Testing Regimen

The Dutch vehicle authority, RDW, granted Tesla a provisional type approval under UN R-171 (Driver Control Assistance Systems) on April 10, 2026, making the Netherlands the first European Union country to authorize the system on public roads . The approval did not come quickly. According to the RDW, the system was "extensively examined and tested for more than one and a half years on our test track and on public roads" . Multiple reports across the automotive press cite more than 18 months of testing, roughly 3,000 hours on a closed test track, 4,500 test scenarios, and approximately 1.6 to 1.8 million kilometers driven on European roads .

The approval remains provisional and currently applies only to vehicles equipped with Tesla's Hardware 4 (AI4) computer in the Netherlands. The RDW is pursuing EU-wide authorization on Tesla's behalf, which will require a vote from member states representing 55% of the EU's population .

The Minister's Defense: Independent Verification

Facing questions in parliament from Labour (PvdA) MP Habtamu de Hoop, Minister Karremans drew a bright line between Tesla's statistical lobbying and the RDW's technical assessment. His key argument was that the disputed safety statistics "played no part" in the final decision; the agency based its approval solely on its own independently verified testing .

Karremans refused to pause or reopen the approval process, and the RDW itself has described the system as making a "positive contribution to safety" . The minister’s stance creates a clear regulatory logic: if a nation's technical agency conducts thousands of hours of its own observation and test-track validation, it renders the manufacturer's self-published claims legally irrelevant.

The European Backdrop: Approving Alone

The Netherlands' solo-first approach has drawn sharp criticism. MP de Hoop pressed the minister on why the country approved the system before any other EU member state, citing fatal crashes in the United States involving Tesla's FSD system as evidence that broader coordination was warranted . The European Transport Safety Council (ETSC) has sent a formal letter to other member states urging them to demand public answers from sources other than Tesla before recognizing the Dutch approval .

Approximately 40,000 Teslas in the Netherlands are now using the FSD (Supervised) feature, according to Tesla Europe, which reported cars using the software had driven 23.6 million kilometers by mid-June . It is important to note that the system is classified as Level 2 driver assistance under UN regulations, meaning the human driver remains legally responsible and must maintain control at all times . Several other EU countries, including Belgium, Denmark, Estonia, and Lithuania, have already recognized the Dutch provisional approval or issued their own preliminary endorsements, creating a complex regulatory patchwork that the European Commission will eventually have to reconcile .