The early results point to a market triumph over bad behavior that lawmakers, police and advocacy groups have struggled to deter. Underwriters who are best able to identify distracted-driving tendencies are better able to price the accompanying risk. Most insurance companies still only offer discounts for good cell-phone behavior, but a growing number are using the findings to raise rates on serial tap-and-swipers.

Distracted driving is more predictive of an eventual loss claim than virtually any other behavior, including speeding and braking. Those who tend to use cellphones at the wheel have 20 percent more insurance claims than others in the risk-pool, according to TrueMotion. “This is creating a pretty clear picture,” said Fiorentino.

Eventually, the margins of underwriters who best identify inattentive drivers will improve, he argued, while their worst customers will get nudged to less vigilant rivals, further muddying those risk pools. As a result, insurance companies of all sizes are hustling to roll out smartphone monitoring systems.

While the proliferation of robot chaperones is encouraging, the data they are collecting is grim. Distraction has increased in every part of the country, despite a rash of new laws intended to curb cellphone use at the wheel. Zendrive said distracted driving levels increased by 10 percent in the past year. Today, the company considers one in 12 drivers a phone addict—on their phones at least one-third of the time at the wheel—and that measure is climbing fast. At the current rate, one in five will fall into that category within three years.

“It’s a massive epidemic,” Zendrive’s Matus said. “And there is clearly a dissonance between what people believe the state of the world is, and what they’re doing.”

Distraction “is embedded in the culture of driving,” said Thomas Dingus, director of the Virginia Tech Transportation Institute. His recent research, which tracks eye patterns, shows drivers are distracted more than half of the time.

Curbing phone use is even a growing problem among self-driving startups. Most of the vehicles in robot-driven fleets still require human chaperones, at least for the time being. That means recruiting and training a safety workforce that has almost nothing to do, if all is going as planned; often, however, it doesn’t. Uber halted its autonomous-car program’s road testing a year ago after one of its vehicles struck and killed a pedestrian. The backup driver in that instance was watching The Voice on her mobile phone at the time.

Ford, which has backed self-driving startup Argo AI, has struggled with driving chaperones falling asleep. Alphabet’s Waymo, widely considered at the front of the self-driving pack, still has two humans in its vehicles much of the time. It turns out people aren’t wired for the kind of “passive vigilance” required to monitor a sentient vehicle. Federal safety regulators have sounded the warning on this, noting that cognitive limitations could be a dangerous stumbling block on the long path from analog driving to fully automated vehicles.

Little of this is surprising, even among the most inattentive drivers. The majority of Americans now consider distracted driving the top threat to safety on the road, far more dangerous than driving under the influence or speeding, according to a recent survey by Volvo.  About one-third of respondents say they turn on “do not disturb” modes on their phones, and three out of four people surveyed would pay more for a vehicle with features designed to prevent distraction.

But it’s now becoming clear that distracted drivers seldom have a good sense of their own conduct. TrueMotion found that those using its monitoring technology reduce their distraction levels by 20 percent after receiving a push notification that identifies their phone use while driving. Those who don’t see their own records of distracted driving, however, use their phones 25 percent more often.