Darcy Bullock, a civil engineering professor at Purdue University, turns to his computer screen to get information about how fast cars are traveling on Interstate 65, which runs 887 miles from Lake Michigan to the Gulf of Mexico. It’s midafternoon on a Monday, and his screen is mostly filled with green dots indicating that traffic is moving along nicely. But near an exit on the outskirts of Indianapolis, an angry red streak shows that cars have stopped moving. A traffic camera nearby reveals the cause: A car has spun out, causing gridlock.
In recent years, vehicles that have wireless connectivity have become a crucial source of information for transportation departments and for academics who study traffic patterns. The data these vehicles emit – including speed, how hard they brake and accelerate, and even if their windshield wipers are on – can offer insights into dangerous road conditions, congestion or poorly timed traffic signals.
“Our cars know more about our roads than agencies do,” said Bullock, who regularly works with the Indiana Department of Transportation to conduct studies on how to reduce traffic congestion and increase road safety. He credits connected-car data with detecting hazards that would have taken years – and many accidents – to find in the past.
The data comes primarily from commercial trucks and from cars made by General Motors that are enrolled in OnStar, GM’s internet-connected service. (Drivers know OnStar as the service that allows them to lock their vehicles from a smartphone app or find them if they have been stolen.) Federal safety guidelines require commercial truck drivers to be routinely monitored, but people driving GM vehicles may be surprised to know that their data is being collected, though it is indicated in the fine print of the company’s privacy policy.
For years, GM has been selling driving data from its cars to third parties. A New York Times report in March described how GM sold information about individuals’ driving behavior to the insurance industry.
After that report, GM was sued by drivers and by the attorney general of Texas for selling consumers’ data without their consent, and the company stopped selling driving data to risk-profiling companies. But it continues to sell anonymous data about where and how its cars are driven, which the company stated was a “common business practice.”
In September, GM released a privacy policy that noted that precise geolocation information and driving behavior for cars enrolled in OnStar might be shared with “affiliates or third parties for research and development purposes (such as university research institutes for improving highway safety).”
‘A wealth of highly desirable information’
For researchers, the connected-car data is a “game changer,” Bullock said. GM’s cars are common enough on America’s roads that the data from them alone is enough to make reliable observations about general trends, he said.
Unlike the data that was sold to the insurance industry, the data that researchers see is anonymized, with no identifying information linked to a particular car. Salman Ahmad, a traffic engineer and researcher, said, however, that it was possible to track a car’s route from start to end in a data set he worked with. He analyzed the driving behavior of people fleeing wildfires, a study that, before connected-car data, would have required after-the-fact surveys of evacuees, whose memories might not have been reliable.
“This is how it actually happened,” Ahmad said. He and his colleagues were able to identify specific intersections and junctions where traffic became snarled. In the future, emergency workers could be sent to those locations during an evacuation to keep traffic moving. Longer term, city planners could consider widening the roads in those areas, or creating alternative routes.
Ahmad said he and his colleagues had purchased the data from Wejo, a connected-car company based in Britain, for about $7,000. Wejo, founded in 2013, offered location and sensor data from nearly 14 million cars, according to an investor presentation from 2023. Former employees said a vast majority of the data had come from GM, which had invested $25 million in the company.
Wejo was beloved by transportation planners for its unique data set. Analysts at the Federal Highway Safety Administration looked at Wejo data on driving behavior and road conditions and found “a wealth of highly desired information,” including “unprecedented” data on when, and for how long, travelers wore seat belts. (Among the findings described in a report released in January: The people least likely to be buckled were those traveling at 120 mph.)
Then, in 2023, Wejo declared bankruptcy.
Do drivers know where their data goes?
The auto industry has been trying to figure out how to make money on data from connected cars, said Roger Lanctot, an industry analyst at StrategiaNow. But the struggles of Wejo, as well as those of an Israeli connected-car company, Otonomo, which was acquired last year after its market value collapsed, suggest that the data is difficult to monetize.
There are also privacy concerns, said Jen Caltrider, a consumer advocate who researched car privacy policies at the Mozilla Foundation. She argues that third-party companies should not get access to the rich trove of data from the myriad sensors in consumers’ cars without an explicit request for permission.
Even if the data is being used for the public good, Caltrider argues, drivers should know it is being collected and have an easy way to stop it from happening, because even if it is anonymous, how and where people drive is sensitive information.
“Anonymized data is never anonymous, especially when location is included,” Caltrider said. “Cars are just so tricky because people aren’t used to getting in their car and thinking they need to play with the settings to protect their privacy.”
This year, General Motors began providing the data it had previously supplied to Wejo to Jacobs Engineering Group, a public company based in Texas, according to a lawsuit filed against GM by the attorney general of Texas. Researchers can gain access to it through StreetLight Data, a transportation analysis subsidiary of Jacobs, Bullock and other researchers said.
“We use connected-vehicle data and work with many departments of transportation all over the country,” said Sean Co, a senior director at StreetLight Data. Jacobs, its parent company, did not respond to requests for comment.
Asked about the sale of information from consumers’ cars, a GM spokesperson pointed back to a statement the company had previously made: “As is common business practice, we share de-identified data not associated with specific drivers or vehicles with select partners to enhance city infrastructure and road safety for pedestrians, cyclists and drivers.” (The Alliance for Automotive Innovation, a trade group, said it did not have enough information to say whether it was a common business practice. A Ford Motor spokesperson said the company did not sell connected-car data to third-party brokers. Other automakers did not respond.)
Asked how consumers could opt out, the GM spokesperson said drivers could cancel OnStar or, for vehicles from 2019 or later, turn off “location services” in the car’s infotainment center settings.
Data as a public ervice
Nathan Sturdevant, the emerging mobility director at the Indiana Department of Transportation, said his agency was still primarily working with data collected through more traditional methods, such as automated car-counting stations installed on the roads, which capture far less granular information than what connected cars provide.
His agency had planned to start ingesting data from Wejo before the company went bankrupt. It is now looking into StreetLight Data, which he said was working on “fuzzifying” the data so that it was not possible to see a car’s most frequently visited locations.
“We don’t care where you’re going and what you’re doing,” Sturdevant said. “It’s just that the vehicles on the road driving all the time is a ton of data and we’d love to have those insights.”
Indiana has made changes based on insights gleaned by Bullock. He looked at places on highways where construction was happening and examined hard braking behavior — when people slam on the brakes as opposed to gently slowing down.
He discovered that a large digital sign warning of work ahead on the back of what is called a “queuing truck” was far more effective at safely slowing traffic than the standard orange diamond signs were, and resulted in an 80% reduction in hard braking. So the agency is now deploying those trucks for any long-term construction projects.
Possible future use of connected-car data, Sturdevant said, would be examining tire pressure monitoring data to identify potholes and pavement conditions.
Sturdevant said that all automakers collected data from connected cars, but that GM had been on the leading edge of monetizing it in the United States. Other automakers, he said, are now hesitant to share data with researchers and government agencies, and are waiting to see the outcome of the lawsuits against GM. “Folks are concerned,” he said.
