According to Tesla’s own calculations, its robotaxis are four times worse at driving than humans, with redactions hiding even more details

By nature, humans are imperfect. Place them behind the wheel of a two-ton vehicle, and various issues can arise. They may collide with stationary objects while driving straight. They can reverse into fixed objects. They hit trees, poles, buses, and trucks. In short, humans lack the rapid reaction times and precise motor skills required to operate a vehicle safely.

However, there’s an issue: all the scenarios described above occurred with Tesla’s autopilot robotaxis. All within a single city and over a period of less than a month. And based on Tesla’s own published data, the company’s autopilot performs four times worse than human drivers.  

A recent “” from indicates that its autonomous robotaxis were involved in five more crashes in Austin, including a collision with a bus while the Tesla was stationary. A Tesla reversed into stationary objects twice: once while moving backward at 2 mph and another time at 1 mph. Another incident involved a collision with a heavy truck at 4 mph, but perhaps the most concerning was a crash into a fixed object at 17 mph while driving straight. 

This is the bulk of what is known, as unlike its autonomous vehicle competitors Waymo and Zoox—and every other market player—Tesla is the sole company to fully redact and conceal crash details from the public, thanks to a confidentiality clause under the NHTSA. Tesla even updated a report for a July 2025 crash—a right-turn collision with an SUV at 2 mph—to note that a victim was hospitalized, reported. The initial report listed only “property damage.” 

Still, the crash disclosures reinforce what Tesla has long claimed about its autopilot program: humans are indeed safer. In fact, following this month’s incidents, they are four times safer according to Tesla’s own metrics. 

By the National Highway Traffic Safety Administration’s standards, Tesla’s autopilot performs even more poorly. Here’s additional information about Tesla’s latest autopilot safety record—or the lack thereof.

The Model Y crash record

The five crashes took place between December 2025 and January 2026, all involving Model Y vehicles with autonomous driving systems active. This accounts for more than a third of all Tesla robotaxi crashes in Austin since the company expanded service to the city last June. 

The fleet logged approximately 700,000 paid miles through November, and is estimated to have exceeded 800,000 miles by mid-January. With 14 crashes since service launched, this equates to roughly one crash every 57,000 miles.

Tesla’s [data] states that the average American driver (of a Tesla) experiences a minor collision every 229,000 miles. This means that over the estimated 800,000 miles driven by Tesla’s vehicles in Austin, the average human driver would have been involved in about four crashes, compared to 14 involving Tesla robotaxis. 

NHTSA guidelines estimate that the average American driver crashes every 500,000 miles—or 1.6 times over the 800,000 miles driven by Tesla’s robotaxis in Austin. Tesla’s fleet crashed at an average rate eight times higher than human drivers. 

Tesla’s Car Troubles

Dan O’Dowd, founder of The Dawn Project, a software industry watchdog, criticized the report’s concealment, stating Tesla was “terrified of the public learning how defective its software is.” Companies are required to report all crashes related to their autonomous and advanced driving systems to the NHTSA within five days. But under NHTSA confidentiality rules, Tesla can simply report that crashes occurred and omit narrative details, instead claiming the details contain “confidential business information.”

Tesla has not responded to comment requests.

In contrast, Waymo’s [data], logging over 127 million miles of autonomous driving, tells a different story that aligns with the narrative that autonomous driving is safer than human driving. The fleet reduced injury-causing crashes by 80% and serious-injury crashes by 91%. The company’s autonomous vehicles have logged 6.34 million miles in Austin through September 2025.

Tesla’s overall driving record is far better—according to their Q3 2025 [report] released in October, the company noted that humans riding in cars with full autopilot engagement were involved in one crash every 6.36 million miles. The company compared this to NHTSA data on non-fully self-driving cars. Last year’s second quarter was their best to date, with one crash every 6.69 million miles for drivers using autopilot. Tesla drivers without autopilot still had fewer crashes than NHTSA figures, at about one crash every 963,000 miles.

The automaker has also faced scrutiny over safety issues with its Cybertruck model, which has exposed numerous design flaws and safety malfunctions. A YouTuber, aiming to demonstrate the Cybertruck’s unsafe front trunk (dubbed a “frunk”), inserted his finger into the gap, resulting in a painful bruise on his finger—and a dent in the trunk. He also inserted a carrot and banana, which the frunk sliced cleanly. 

The Cybertruck features a $61,000 “armor glass” window that was quickly [compromised]. The parents of a 19-year-old are suing the automaker after [an incident], alleging the car’s door handles made escape impossible. The NHTSA is also investigating Tesla’s Model 3 sedans and [other models] as a major safety concern. The Model Y is under review for its door handles following NHTSA reports of people being trapped due to low battery voltage. Though there are manual handles inside, children trapped may not reach them—a growing concern after multiple reports of entrapment in [vehicles]. Just last year, a jury found Tesla partially responsible for a crash involving its Autopilot technology and ordered the company to [pay] in damages. 

Elon Musk is vocal about Tesla’s legal battles over design flaws. After the NHTSA [acted on] autonomous vehicles in 2023, the billionaire founder took to [social media] to [argue] that the word “recall” was “anachronistic and just flat-out wrong!”