Tesla released its Autopilot safety report for Q2 2025, confirming that Autopilot safety has regressed so far in 2025.
The Autopilot safety reports consist of Tesla releasing the miles driven between crashes for Tesla vehicles with Autopilot features turned on, and comparing that with the miles driven by vehicles with Autopilot technology with the features not turned on, as well as the US average mileage between crashes.
We have often highlighted how Tesla’s Autopilot safety report is flawed in many ways:
- Methodology is self‑reported. Tesla counts only crashes that trigger an airbag or restraint; minor bumps are excluded, and raw crash counts or VMT are not disclosed.
- Road type bias. Autopilot is mainly used on limited‑access highways—already the safest roads—while the federal baseline blends all road classes. Meaning there are more crashes per mile on city streets than highways.
- Driver mix & fleet age. Tesla drivers skew newer‑vehicle, higher‑income, and tech‑enthusiast; these demographics typically crash less.
The primary value of Tesla’s quarterly Autopilot safety reports lies in comparing the miles between crashes with Autopilot features turned on over time.
Even then, it remains problematic, as Tesla stopped reporting the data for over a year. When it resumed reporting last year, it edited the previously released data.
Today, Tesla released its Autopilot safety report for Q2 2025:
The data clearly shows that mileage between accidents has shorten with Autopilot engaged in 2025 compared to the same period last year:
| Metric (miles between crashes; higher = safer) | Q1 2024 | Q1 2025 | YoY Δ Q1 | Q2 2024 | Q2 2025 | YoY Δ Q2 |
|---|---|---|---|---|---|---|
| Autopilot engaged | 7.63 M | 7.44 M | −2.5 % | 6.88 M | 6.69 M | −2.8 % |
| No Autopilot (active‑safety only) | 0.955 M | 1.51 M | +58.1 % | 1.45 M | 0.963 M | −33.6 % |
| U.S. fleet average (NHTSA/FHWA) | 0.670 M | 0.702 M | +4.8 % | 0.670 M | 0.702 M | +4.8 % |
This is happening amid Tesla being challenged in California court over false advertising of its Autopilot and Full Self-Driving features.
Electrek’s Take
Again, it’s important to put this data into context. Tesla releases this positively by comparing Autopilot-engaged mileage between crashes to the US fleet average.
It uses this data to claim it is “10x safer than human driving”, but this is highly misleading.
Top comment by Grant
I'm not a fan of the cherry-picked data Tesla uses. This is significant. Also, one coder put it this way "It's really easy to get cars to drive themselves in 95% of conditions, it's the last 5% that is really tricky." Isn't that the rub with Tesla (and others). It makes it look like they are almost ready indefinitely.
First off, it’s not Autopilot versus humans. It’s Autopilot plus humans against humans. And that’s on top of all the other issues with self-reporting, highway driving versus city driving, etc.
Therefore, the only value of this report is comparing Autopilot-engaged mileage over time, and it clearly regressed in 2025.
There’s no doubt about that.
It is particularly noteworthy coming from Tesla, which has consistently refused to release any data about its self-driving efforts. This is the only data it is releasing, and it is extremely limited, yet it indicates that Autopilot safety regressed in 2025.
FTC: We use income earning auto affiliate links. More.
Comments