When Insurance Admits AI is Safer Than People
When better data forces markets to reprice reality
A few weeks ago I wrote about how quickly my family adapted to Tesla’s Full Self-Driving capability, to the point where my seven-year-old now asks “Why isn’t the car driving itself?” when we’re not in FSD mode. The post sparked more replies and texts than I expected, with both friends and family convinced to try it, a few shopping for Teslas, and perhaps most important for my life, my wife now being comfortable using it. The technology, it seems, has crossed some psychological threshold.
Today, that psychological threshold got a financial stamp of approval.
Insurance company Lemonade announced they’re launching Autonomous Car Insurance, cutting premiums by roughly 50% when FSD is engaged. From the announcement:
The new offering cuts per-mile rates for FSD-engaged driving by approximately 50%, reflecting what the data shows to be significantly reduced risk during autonomous operation. Lemonade expects further reductions as Tesla releases FSD software updates, which are anticipated to make the cars even safer over time.
And later:
“Teslas driven with FSD are involved in far fewer accidents,” [Lemonade President Shai] Wininger added. “By connecting to the Tesla onboard computer, our models are able to ingest incredibly nuanced sensor data that lets us price our insurance with higher precision than ever before.”
“Beyond the product announcement today, we’re also announcing our commitment to the Tesla community – the safer FSD software becomes, the more our prices will drop,” Wininger said.
I admit I reread the price decrease a few times to confirm. This is not a 5% drop. It’s fifty percent. The data is so compelling that Lemonade is willing to take in less in premiums because that's how confident they are that they will have to pay out less in losses. Talk about an amazing case study of how AI and robotics deployment within one category can reshape adjacent industries.
First Mover Advantage
Insurance is fundamentally about pricing risk. The better you can predict outcomes, the more precisely you can price coverage. Lemonade is betting real money that FSD miles are safer than human miles - safe enough to cut rates in half while still maintaining profitable loss ratios.
This isn’t just a new product announcement, it’s a very competitive move. Traditional insurers are grappling with how to price coverage in an increasingly autonomous world. Lemonade is using AI to price AI and believes they will lead the industry going forward.
Now, committing to lower prices further as Tesla releases FSD updates that improve safety can be taken as a brilliant customer acquisition gamble versus genuine confidence in the data. My guess is it’s both. Other insurers will watch closely. If Lemonade starts to attract customers and loss ratios hold or improve, expect to see this roll out more broadly across the industry as traditional underwriting models built on driver age, ZIP code, and proxies start to look obsolete.
An Important Milestone
It is still early in the autonomous driving world. There are between 5 and 6 million Teslas on the road with FSD capabilities, FSD still requires driver supervision, and full autonomy is still probably years away. Even Autonomous Car Insurance will only be available in Arizona and then Oregon to start.
But milestones matter. And an insurance company pricing autonomous miles at half the rate of human miles is a meaningful one toward an increasingly AI-driven (pun intended) future.
Repricing Reality
The question my son asks about the car not driving itself applies more broadly: why isn’t [insert industry] doing this yet? Often, the constraint isn’t capability. It’s the gap between what technology enables and what we’re willing to let it do.
But when we deploy AI and robotics into industries where humans currently operate, adjacent systems and supporting industries have to adapt. Insurance is an obvious example. But across manufacturing, healthcare, logistics, food service - any industry with physical human risk - will need to rethink how they price, regulate, and organize around more intelligent machines.
We will need to reprice reality.
Tipping Points
Volvo just unveiled their new EX60 electric SUV. Impressive range, fast charging, premium interior. But only ‘driver assist’ features, no self-drive.
At this point, that’s a dealbreaker for me (and I’m guessing for a large and growing number of drivers). Should insurance prices continue to fall with autonomy, “driver assist only” stops being a neutral product decision and becomes a tax. And when the absence of a technology becomes more notable than its presence, you’re past the tipping point.
N.B. In another example of regulation being behind the times, according to Lemonade California will likely not see this product in market for a while thanks to Prop. 103. Passed in 1988, this piece of legislation, among other things, prohibits insurance companies from pricing insurance based on telemetry data. Ironically (?) it was originally passed to protect consumers, lower costs and create a better insurance market 🤦🏽♂️