A new Tesla safety experiment is blurring the line between emergency intervention and assisted driving. The company says Full Self-Driving (Supervised) v14.3.9 will include an active-safety capability that can engage when an imminent collision is detected and ordinary Automatic Emergency Braking may not be enough. Separate reporting on internal release notes describes the feature as Automatic Collision Evasion and says it can also step in when the driver appears heavily distracted or when FSD may have been unintentionally disengaged.
The idea is simple but consequential: instead of only warning or braking, the car could briefly take broader control to steer, brake and manage speed through a dangerous moment. Tesla is still testing the update, and important questions about rollout, eligibility, driver control and real-world reliability remain.
A Safety Net That Can Wake Up While the Driver Is in Control
Tesla’s Automatic Collision Evasion feature is designed to operate even when a person is driving manually. Tesla said FSD Supervised v14.3.9 can activate on the driver’s behalf when the system detects an imminent collision and determines that Automatic Emergency Braking may not be sufficient. Reports based on internal release notes say employees are testing the feature as part of software update 2026.27.5 before a broader public release.
That changes the relationship between FSD and the driver. Normally, FSD Supervised is something the driver chooses to engage. Under this concept, parts of the driving system could remain available in the background as an emergency layer. Tesla also said intervention may occur when severe distraction is detected or when FSD appears to have been disengaged unintentionally. In practical terms, a manual drive could suddenly become an assisted one if the software judges that collision risk has become urgent.
Why Braking Alone Is Not Always Enough
Automatic Emergency Braking is built around a narrow task: detect an imminent crash and apply the brakes when the driver has not reacted adequately. Tesla’s manuals describe AEB as a system intended to reduce impact severity rather than guarantee collision avoidance. It can operate across a wide speed range, but Tesla also warns that road debris, partial obstructions, weather and other factors can produce missed, delayed or inappropriate braking.
Automatic Collision Evasion is intended to go one step further. If stopping distance alone cannot resolve the danger, a safe response may require steering around an obstacle, changing the vehicle’s path or combining steering with braking. That is a harder problem because the car must judge not only what is ahead, but also whether an escape path is clear. A maneuver that avoids one object can create another risk in an adjacent lane, shoulder or intersection, making accurate scene understanding critical.
Driver Attention Becomes Part of the Trigger
Tesla uses an in-cabin camera to monitor attentiveness while FSD Supervised is engaged. Its owner manuals say the camera watches whether the driver remains engaged with the road and available to take control. Repeated looks away can trigger warnings, while ignored prompts can disable assisted-driving functions for that drive. The monitoring system remains active even when a driver is wearing sunglasses.
The new safety concept potentially gives that attention signal a more active role. Tesla has said FSD may engage if the vehicle detects that the driver is heavily distracted, rather than merely issuing another warning. Consider a driver reaching toward the rear seat as traffic suddenly compresses ahead: the system could use both roadway perception and cabin monitoring to decide whether intervention is warranted. That approach may shorten reaction time, but it also makes the accuracy of distraction detection important because a mistaken intervention could surprise an attentive driver.
FSD Still Does Not Become Autonomous
The presence of an automatic emergency takeover does not change Tesla’s description of Full Self-Driving Supervised. Tesla says the system can steer, accelerate, brake, change lanes, negotiate intersections and navigate on many road types, but it still requires active driver supervision. The company explicitly states that FSD Supervised does not make the vehicle autonomous and that the person behind the wheel remains responsible for control.
NHTSA uses the same category. In a 2025 investigation, NHTSA noted that Tesla characterizes FSD as SAE Level 2 partial automation, which means the driver must actively supervise the driving task and intervene when necessary. That distinction matters during an emergency activation. Even if the car initiates evasive action on its own, the driver is not transformed into a passenger. The difficult human-factors question is whether someone can immediately understand what the vehicle is doing and safely counter it if the automated response is wrong.
Tesla’s Camera-Based Approach Faces a Hard Test
FSD builds its understanding of the road primarily from cameras and neural-network processing. Tesla’s manuals say exterior cameras around the vehicle create a model of the environment, which the onboard computer uses to make decisions. That architecture allows the same perception stack to support lane changes, turns, obstacle avoidance and, potentially, emergency evasive action without adding a sensor suite for the feature.
The limitation is that cameras can be degraded by conditions that also challenge human vision. In March 2026, NHTSA upgraded an investigation into about 3.2 million FSD-equipped Teslas after examining crashes in reduced-visibility situations such as glare and airborne obscurants. The agency said some cases involved insufficient detection of degraded camera conditions or alerts that came too late for the driver to respond. Automatic Collision Evasion arrives in an environment where fast, accurate perception is not merely useful; it is the foundation of the emergency intervention itself.
There Is Strong Evidence That Automatic Braking Saves Crashes
The safety case for automated intervention has years of evidence behind it. Research shows that automatic emergency braking can reduce crash types. The Insurance Institute for Highway Safety reports that front AEB cuts rear-end crash rates by about 50 percent and rear-end crashes involving injuries by 56 percent. Those findings helped establish automatic braking as a driver-assistance technology.
Federal policy reflects that evidence. NHTSA finalized a rule requiring AEB, including pedestrian detection, on new passenger cars and light trucks by September 2029. The agency projected the standard would save at least 360 lives and prevent at least 24,000 injuries each year. Tesla’s feature is attempting to address the situations left beyond that protection—moments when braking alone may not be enough. The potential benefit is clear, but steering intervention introduces more variables than straight-line braking, so performance will need to be judged separately rather than assumed from AEB’s record.
Tesla’s Safety Numbers Are Encouraging but Need Context
Tesla argues that FSD Supervised already lowers collision risk. Its Vehicle Safety Report says Teslas using FSD experience seven times fewer major and minor collisions and five times fewer off-highway collisions than its comparison groups. In Europe, Tesla reported that supervised FSD had 4.1 times fewer collisions than manually driven Teslas across more than 100 million kilometres in five countries.
Those figures matter, but they are not the same as an independent randomized safety trial. Driving environment, road type, vehicle mix, driver behavior and the circumstances in which people choose to activate FSD can affect comparisons. Reuters has noted skepticism over the validity of Tesla’s safety comparisons and how those figures should be interpreted overall. For Automatic Collision Evasion, the most useful evidence will be specific: how often it activates, how many crashes it prevents, how often drivers override it, and whether false or unnecessary interventions create new hazards.
The Feature Arrives Under Active Federal Scrutiny
Tesla is introducing this capability while U.S. regulators are examining FSD performance in several areas. NHTSA’s Engineering Analysis into reduced roadway visibility covers 3.2 million vehicles and focuses on whether FSD can recognize visibility degradation and warn drivers with enough time to respond. Separately, a 2025 preliminary evaluation covers 2.88 million FSD-equipped vehicles and examines traffic-law violations, including red-light behavior and movement into opposing lanes.
Those investigations do not establish that an FSD-equipped Tesla is defective, but they show why an automatically triggered version of the system will receive attention. A feature that chooses to take control without a driver command raises questions about activation thresholds, warning timing and the driver’s opportunity to intervene. NHTSA’s traffic-violation probe specifically considers whether unexpected FSD inputs can leave a driver too little time to supervise safely. Those same issues become central when FSD is being used as a crash-avoidance backstop.
Who Gets It, and When, Is Still Unclear
Tesla has said v14.3.9 is entering rollout, but reporting on software update 2026.27.5 indicates Automatic Collision Evasion was still in employee testing when the feature surfaced. Tesla often stages software releases, exposing a limited group of vehicles before expanding availability. A feature can also be delayed or modified if testing reveals problems, so an announcement of imminent rollout is not the same as universal availability.
Eligibility may be a dividing line. Reporting on the notes says the capability is tied to FSD access, while Tesla’s support pages emphasize that FSD availability varies by hardware, software version, region, model, trim and model year. Tesla offers v14 trials on Model S, Model 3, Model X, Model Y and Cybertruck vehicles in North American markets. Owners should not assume that every Tesla with active-safety features will receive Automatic Collision Evasion at the same time—or receive it at all.
The Bigger Shift Is From Warning Drivers to Sharing the Rescue
Automatic Collision Evasion fits an industry move toward combining safety systems rather than treating them as features. In 2026, the National Transportation Safety Board recommended that partial-automation systems integrate safety-critical technologies such as automatic emergency braking and driver monitoring, while addressing automation complacency and misuse. NHTSA has expanded its testing of advanced driver-assistance systems, with the 2026 Model Y becoming the first vehicle to pass its newer ADAS benchmark.
Tesla’s test pushes that integration further by allowing the driving stack to become an emergency responder. If it works reliably, the benefit could be measured in the fractions of a second that separate a near miss from a serious crash. If it misjudges a scene, however, the intervention can be more consequential than a warning tone. The feature’s significance will depend less on its name than on data showing when it activates, what it does, and whether drivers are safer afterward.
