Uber is reportedly planning to bring to San Francisco an autonomous ride-sharing program similar to its pilot program launched in Pittsburgh earlier this year where customers using Uber’s app can be picked up by an “autonomous vehicle” with a company engineer behind the wheel. The company wrote in a press release:

Starting today, riders who request an uberX in San Francisco will be matched with a Self-Driving Uber if one is available. Expanding our self-driving pilot allows us to continue to improve our technology through real-world operations. With its challenging roads and often varied weather, Pittsburgh provided a wide array of experiences. San Francisco comes with its own nuances including more bikes on the road, high traffic density and narrow lanes.

The problem is that Uber doesn’t have a permit to operate autonomous vehicles in California and it justifies its right to operate anyway through some sketchy semantics.

We reported earlier this week that NVIDIA became the 20th company to be allowed to test autonomous vehicles on California’s public roads. GM, Ford, Google, Tesla and others are on that list, but Uber is not, and apparently they don’t plan to be.

It looks like the company wants its autonomous technology to be considered a level 3, which requires active monitoring by a driver when activated and would comply with the law since California’s DMV defines autonomous systems as able to drive “without the active physical control or monitoring by a human operator”.

But apparently, Uber still wants to refer to the system as “self-driving.”

The company’s self-driving lead engineer, Anthony Levandowski, told Financial Times that “the fleet would be “self-driving” but not “autonomous”, making an unusual semantic distinction.” He added:

“It is effectively not good enough to be an autonomous vehicle yet. Our technology is not capable of driving without human monitoring or active physical control.”

Through the distinction, Uber will not be required to publicly release safety statistics like collisions reports or disengagement reports.

Levandowski told FT:

“I don’t know that we are going to plan on releasing that [collision data]; it is proprietary information. We do want to be transparent. We might share other statistics that might be more relevant.”

Either Uber’s autonomous system is really just capable of level 3 autonomy and needs constant monitoring from a driver and in that case, it’s strange that the company insists on calling it “self-driving,” or it’s another example of the company pushing the boundaries of regulations, as it has done on several occasions in various markets for its ride-sharing service.