You may have seen this story making the rounds: Tesla drivers in China are buying $30 plastic doll heads and mounting them near the cabin camera to fool the driver monitoring system. One guy reportedly drove 30 minutes on the highway — one hand in a bag of sunflower seeds, the other holding a phone — while a fake Dwayne Johnson head convinced the car everything was fine.

It's a funny image. It's also a perfect example of what happens when technology outpaces the policies meant to govern it — and in commercial fleets, that gap is wider than most operators realize.

ADAS Is Already in Your Fleet

Advanced driver assistance features (adaptive cruise control, lane-keeping assist, automatic emergency braking, hands-free highway driving) are no longer optional upgrades on the vehicles rolling into fleet spec sheets. They're standard.

That means your drivers are already behind the wheel of vehicles that, in certain conditions, will steer, brake, and accelerate on their own. Some of those drivers understand exactly what those systems can and can't do. Others don't.

And here's the uncomfortable truth: the better these systems get, the more drivers treat them as a reason to disengage. Not out of malice — out of the entirely human tendency to trust something that keeps working. The Tesla doll-head story is an extreme example of that tendency. But the underlying behavior — a driver mentally checking out because the car "has it" happens at a much lower threshold than a fake celebrity figurine.

A Single Camera Was Never Going to Be Enough

The reason a $30 toy could defeat Tesla's monitoring system is that the system was trying to infer whether a human is actually paying attention from a single input: a forward-facing camera looking for a head in roughly the right position.

That's not a knock specific to Tesla. It illustrates something fleet operators should internalize more broadly: no single data point tells you whether a driver is engaged.

This is why fleet telematics worth its price tag paints a behavioral picture — hard braking events, sudden acceleration, following distance, distracted driving flags, hours-of-service patterns. Taken together, those data points surface something a cabin camera alone can't: a driver's actual habits behind the wheel, over time, across conditions.

But even strong telematics is largely retrospective. It tells you what happened. It doesn't prevent a driver from negative behaviors.

That's where policy has to do the work that technology can't.

Most Fleet ADAS Policies Don't Exist Yet

When was your fleet vehicle use policy last updated? If the answer is more than a few years ago, there's a reasonable chance it was written before lane-keeping assist was common equipment — let alone before hands-free highway systems entered the picture.

Most fleet policies cover the basics well: seatbelts, speed limits, phone restrictions, accident reporting procedures. Very few address how drivers are expected to interact with driver-assist features while operating a company vehicle.

A meaningful ADAS policy needs to answer a handful of questions:

• Which features are approved for use on the job, and under what road or traffic conditions?
• What does "driver responsibility" mean when those features are active? (Spoiler: it means everything it always did.)
• What's explicitly off the table — phone use, eating, anything that pulls attention away while assist features are running?
• What are the consequences for deliberately defeating or ignoring safety systems?

That last one matters more than it might seem. If a driver in your fleet finds a way to game a monitoring system — or simply ignores it repeatedly — that's not a technology failure you can blame on the OEM. It's a conduct issue, and your policy needs to name it as one.

The Accountability Doesn't Live in the Car

NHTSA currently has an engineering analysis open covering 3.2 million Tesla vehicles, prompted in part by documented FSD violations including running red lights and crossing into oncoming lanes. Every incident in that investigation shares a common thread: a system designed to assume human supervision, operating without it. When something goes wrong with a company vehicle, the question won't stop at what the driver did. It will extend to what the organization knew about how those vehicles were being operated, what policies governed that operation, and what training supported those policies.

The technology in your fleet is getting more capable every year. Driver behavior, left unaddressed, tends to expand to fill whatever space that capability creates. Policy is how you define the space.

A plastic Dwayne Johnson head is an absurd solution to a problem drivers shouldn't be trying to solve. But the instinct behind it — to let the car handle it — isn't going away. Fleet managers who get ahead of that instinct with clear policy and strong oversight will be in a much better position than those who assume the technology is handling it.

It isn't. Not yet. Maybe not ever.