Tesla’s latest 2025.44 over-the-air software update has been making headlines on the internet. Developers and self-driving enthusiasts have identified hidden code that plots several geofenced areas, which suggests that Tesla may be about to launch an unsupervised Full Self-Driving (FSD) mode in its customer fleet.
One of the most prominent entries on the code is named Bay Unsupervised CA DMV, and it encompasses a large portion of the San Francisco Bay Area. There are also smaller, more specific areas, such as the parking lot of Tesla Engineering Headquarters in Palo Alto, and larger areas, such as a whole neighborhood of Palo Alto and San Francisco.
Why These Geofences Are Significant
Geofencing is a significant technology in self-driving. Geofencing will enable companies such as Tesla to restrict the range of self-driving systems, preferably to locations where they are most confident in their effectiveness and safety. More to the point, the naming scheme of these zones actively implies that Tesla is not merely testing, but intends to switch on unsupervised FSD, i.e., driving without a safety driver or human supervision in specific designated areas.
Tesla adds "Bay Unsupervised CA DMV" geofence to customer vehicles in latest software update
byu/YeetYoot-69 inSelfDrivingCars
“L3/L4 geofence preparations for customer cars appear to be in the works now … My best guess is that they’re going to roll out level 3 operation under some conditions to customer vehicles.”
Tesla has already initiated FSD-monitored ride-hailing pilot programs among the employees in the Bay Area and Austin. These trials involve the supervised operation of the robotaxi, implying that the initial approach is to test and validate within limited areas, then proceed with expansion.
In the meantime, Tesla CEO Elon Musk has already talked about unsupervised FSD, which he called an unsupervised localized parameter set, limited to safe, mapped zones in the beginning.
It is also mentioned that Tesla is working towards integrating its supervised FSD stack with the unsupervised (robotaxi) stack, which could form a single neural model. Assuming success, this would significantly accelerate Tesla’s autonomy roadmap by enabling robotaxi-level driving in customer vehicles, albeit under limited conditions, such as geofenced areas.
Risks, Challenges, and Regulatory Hurdles
However, there are some concerns about this possible rollout. Other observers are still sceptical. Indicatively, according to Electrek, Tesla has, to date, rolled out its unsupervised system only within geofenced areas and relies extensively on teleoperation, which implies that the scalability of the system is questionable.
Another significant obstacle might be regulators: the uncontrolled driving in its entirety has legal and insurance obstacles, particularly on accountability, safety level, and control.
Users have their doubts as well: Will unsupervised FSD just imply that the car will no longer nag the driver – or will it really drive by itself? One of the comments made on Reddit states:
“My prediction is that they’ll define ‘Unsupervised’ … rather than the driver not supervising the car … the car won’t nag if you read a book … but the driver still pays the fall for failures.”
