US safety agency pressures Texas-based Tesla over full self-driving claims, crash concerns

The U.S. National Highway Traffic Safety Administration has raised concerns about Tesla's public messaging on its "Full Self-Driving" system. Photo via tesla.com

The U.S. government's highway safety agency says Tesla is telling drivers in public statements that its vehicles can drive themselves, conflicting with owners manuals and briefings with the agency saying the electric vehicles need human supervision.

The National Highway Traffic Safety Administration is asking the company to “revisit its communications” to make sure messages are consistent with user instructions.

The request came in a May email to the company from Gregory Magno, a division chief with the agency's Office of Defects Investigation. It was attached to a letter seeking information on a probe into crashes involving Tesla's “Full Self-Driving” system in low-visibility conditions. The letter was posted Friday on the agency's website.

The agency began the investigation in October after getting reports of four crashes involving “Full Self-Driving" when Teslas encountered sun glare, fog and airborne dust. An Arizona pedestrian was killed in one of the crashes.

Critics, including Transportation Secretary Pete Buttigieg, have long accused Tesla of using deceptive names for its partially automated driving systems, including “Full Self-Driving” and “Autopilot,” both of which have been viewed by owners as fully autonomous.

The letter and email raise further questions about whether Full Self-Driving will be ready for use without human drivers on public roads, as Tesla CEO Elon Musk has predicted. Much of Tesla's stock valuation hinges on the company deploying a fleet of autonomous robotaxis.

Musk, who has promised autonomous vehicles before, said the company plans to have autonomous Models Y and 3 running without human drivers next year. Robotaxis without steering wheels would be available in 2026 starting in California and Texas, he said.

A message was sent Friday seeking comment from Tesla.

In the email, Magno writes that Tesla briefed the agency in April on an offer of a free trial of “Full Self-Driving” and emphasized that the owner's manual, user interface and a YouTube video tell humans that they have to remain vigilant and in full control of their vehicles.

But Magno cited seven posts or reposts by Tesla's account on X, the social media platform owned by Musk, that Magno said indicated that Full Self-Driving is capable of driving itself.

“Tesla's X account has reposted or endorsed postings that exhibit disengaged driver behavior,” Magno wrote. “We believe that Tesla's postings conflict with its stated messaging that the driver is to maintain continued control over the dynamic driving task."

The postings may encourage drivers to see Full Self-Driving, which now has the word “supervised” next to it in Tesla materials, to view the system as a “chauffeur or robotaxi rather than a partial automation/driver assist system that requires persistent attention and intermittent intervention by the driver,” Magno wrote.

On April 11, for instance, Tesla reposted a story about a man who used Full Self-Driving to travel 13 miles (21 kilometers) from his home to an emergency room during a heart attack just after the free trial began on April 1. A version of Full Self-Driving helped the owner "get to the hospital when he needed immediate medical attention,” the post said.

In addition, Tesla says on its website that use of Full Self-Driving and Autopilot without human supervision depends on “achieving reliability" and regulatory approval, Magno wrote. But the statement is accompanied by a video of a man driving on local roads with his hands on his knees, with a statement that, “The person in the driver's seat is only there for legal reasons. He is not doing anything. The car is driving itself,” the email said.

In the letter seeking information on driving in low-visibility conditions, Magno wrote that the investigation will focus on the system's ability to perform in low-visibility conditions caused by “relatively common traffic occurrences.”

Drivers, he wrote, may not be told by the car that they should decide where Full Self-Driving can safely operate or fully understand the capabilities of the system.

“This investigation will consider the adequacy of feedback or information the system provides to drivers to enable them to make a decision in real time when the capability of the system has been exceeded,” Magno wrote.

The letter asks Tesla to describe all visual or audio warnings that drivers get that the system “is unable to detect and respond to any reduced visibility condition.”

The agency gave Tesla until Dec. 18 to respond to the letter, but the company can ask for an extension.

That means the investigation is unlikely to be finished by the time President-elect Donald Trump takes office in January, and Trump has said he would put Musk in charge of a government efficiency commission to audit agencies and eliminate fraud. Musk spent at least $119 million in a campaign to get Trump elected, and Trump has spoken against government regulations.

Auto safety advocates fear that if Musk gains some control over NHTSA, the Full Self-Driving and other investigations into Tesla could be derailed.

Musk even floated the idea of him helping to develop national safety standards for self-driving vehicles.

“Of course the fox wants to build the henhouse,” said Michael Brooks, executive director of the Center for Auto Safety, a nonprofit watchdog group.

He added that he can't think of anyone who would agree that a business mogul should have direct involvement in regulations that affect the mogul’s companies.

“That’s a huge problem for democracy, really,” Brooks said.

Trending News

A View From HETI

A new report predicts that Texas will be home to 30 percent of the U.S. data center market by 2028. Photo via Chevron.com

Data centers are proving to be a massive economic force in Texas.

For instance, a new report from clean energy company Bloom Energy predicts Texas will see a 142 percent increase in its market share for data centers from 2025 to 2028. That would be the highest increase of any state.

Bloom Energy expects Texas to exceed 40 gigawatts of data-center capacity by 2028, representing a nearly 30 percent share of the U.S. market. A typical AI data center consumes 1 to 2 gigawatts of energy.

“Data center and AI factory developers can’t afford delays,” Natalie Sunderland, Bloom Energy’s chief marketing officer, said in the report. “Our analysis and survey results show that they’re moving into power‑advantaged regions where capacity can be secured faster — and increasingly designing campuses to operate independently of the grid.”

“The surge in AI demand creates a clear opportunity for states that can adapt to support large-scale AI deployments at speed,” Sunderland adds.

Further evidence of the data center explosion in Texas comes from ConstructConnect, a provider of data and software for contractors and manufacturers. ConstructConnect reported that in the 12-month span through November 2025, data-center construction starts in Texas accounted for $11 billion in spending. At $12.5 billion, only Louisiana surpassed the Texas total.

Capital expenses for U.S. data centers were expected to surpass $425 billion last year, according to ratings agency S&P Global.

ConstructConnect also reports that Texas is among five states collectively grabbing 80 percent of potential data center construction starts. Currently, Texas hosts around 400 data centers, with close to 60 of them in the Houston market.

A large pool of data-center construction spending in Texas is flowing from Google, which announced in November that it would earmark $40 billion for new AI data centers in the state.

“Texas leads in AI and tech innovation,” Gov. Greg Abbott proclaimed when the Google investment was unveiled.

Other studies and reports lay out just how much data centers are influencing economic growth in the Lone Star State:

  • A study by Texas Royalty Brokers indicates Texas leads the U.S. with 17 clusters of AI data centers. The study measured the density of AI data centers by counting the number of graphics processing units (GPUs) installed in those clusters. GPUs are specialized chips built to run AI models and perform complex calculations.
  • Citing data from construction consulting company FMI, The Wall Street Journal reported that spending on construction of data centers is expected to rise 23 percent in 2026 compared with last year. Much of that construction spending will happen in Texas. In the 12 months through November 2025, the average data center cost $597 million, according to ConstructConnect.
  • Data published in 2025 by commercial real estate services company Cushman & Wakefield shows three Texas markets — Austin, Dallas and San Antonio — boast the lowest construction costs for data centers among the 19 U.S. markets that were analyzed. The mid-range of costs in that trio of markets is roughly $10.65 million per megawatt. Houston isn’t included in the data.

Although Houston isn’t cited in the Cushman & Wakefield data, it nonetheless is playing a major role in the data-center boom. Houston-area energy giants Chevron and ExxonMobil are chasing opportunities to supply natural gas as a power source for data centers, for example.

“As Houston rapidly evolves into a hub for AI, cloud computing, and data infrastructure, the city is experiencing a surge in data-center investments driven by its unique position at the intersection of energy, technology, and innovation,” says the Greater Houston Partnership.

Trending News