Tesla launched its Autopilot security report for Q2 2025, confirming that Autopilot security has regressed up to now in 2025.
The Autopilot security experiences include Tesla releasing the miles pushed between crashes for Tesla automobiles with Autopilot options turned on, and evaluating that with the miles pushed by automobiles with Autopilot know-how with the options not turned on, in addition to the US common mileage between crashes.
Now we have typically highlighted how Tesla’s Autopilot security report is flawed in some ways:
- Methodology is self‑reported. Tesla counts solely crashes that set off an airbag or restraint; minor bumps are excluded, and uncooked crash counts or VMT aren’t disclosed.
- Street kind bias. Autopilot is especially used on restricted‑entry highways—already the most secure roads—whereas the federal baseline blends all street lessons. That means there are extra crashes per mile on metropolis streets than highways.
- Driver combine & fleet age. Tesla drivers skew newer‑automobile, increased‑revenue, and tech‑fanatic; these demographics usually crash much less.
The first worth of Tesla’s quarterly Autopilot security experiences lies in evaluating the miles between crashes with Autopilot options turned on over time.
Even then, it stays problematic, as Tesla stopped reporting the information for over a 12 months. When it resumed reporting final 12 months, it edited the beforehand launched information.
Right this moment, Tesla launched its Autopilot security report for Q2 2025:
The information clearly reveals that mileage between accidents has shorten with Autopilot engaged in 2025 in comparison with the identical interval final 12 months:
| Metric (miles between crashes; increased = 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 (lively‑security solely) | 0.955 M | 1.51 M | +58.1 % | 1.45 M | 0.963 M | −33.6 % |
| U.S. fleet common (NHTSA/FHWA) | 0.670 M | 0.702 M | +4.8 % | 0.670 M | 0.702 M | +4.8 % |
That is taking place amid Tesla being challenged in California courtroom over false promoting of its Autopilot and Full Self-Driving options.
Electrek’s Take
Once more, it’s vital to place this information into context. Tesla releases this positively by evaluating Autopilot-engaged mileage between crashes to the US fleet common.
It makes use of this information to say it’s “10x safer than human driving”, however that is extremely deceptive.
First off, it’s not Autopilot versus people. It’s Autopilot plus people towards people. And that’s on high of all the opposite points with self-reporting, freeway driving versus metropolis driving, and many others.
Subsequently, the one worth of this report is evaluating Autopilot-engaged mileage over time, and it clearly regressed in 2025.
There’s little doubt about that.
It’s significantly noteworthy coming from Tesla, which has persistently refused to launch any information about its self-driving efforts. That is the one information it’s releasing, and this can be very restricted, but it signifies that Autopilot security regressed in 2025.
