Tesla drivers say new FSD update is repeatedly running red lights
Reports that Tesla’s latest Full Self-Driving (FSD) software is attempting to run red lights and ignore school bus stop signs are fueling concern over the system’s safety and Tesla’s decision to ship it widely while keeping detailed performance data opaque. Commenters debate how FSD compares statistically to human drivers and to more constrained systems like Waymo, but note that without transparent, rigorous safety metrics, claims of superiority are not verifiable. The thread also raises legal and ethical questions around marketing an SAE Level 2 system as “Full Self-Driving,” shifting liability to drivers, and normalizing dangerous failure modes in safety‑critical software.
Technical behavior & failures
- Multiple anecdotes of recent FSD versions running red lights, ignoring school bus stop arms, mis-handling stop signs, and in at least one case striking a curb hard enough to damage wheels.
- Some commenters argue that even if no collision occurs, running red lights or entering bike lanes is inherently unsafe.
- Others ask whether FSD is actually clearing red lights only when “safe” (no cross-traffic), but are pushed back on the idea that this can ever be considered acceptable behavior.
Safety vs human drivers & data transparency
- Several participants ask how FSD safety compares statistically to human drivers.
- Repeated point: Tesla does not release the detailed, methodologically clear safety data needed to make this comparison.
- Tesla is accused of cherry‑picking numbers, omitting base rates and confidence intervals, and previously promoting misleading safety stats, whereas Waymo is cited as publishing more rigorous analyses.
- One commenter who has used FSD heavily claims it is “orders of magnitude less safe” than an average human and would be deadly without close supervision; others contest this based on their own positive experiences.
User experiences & expectations
- Some Tesla owners say FSD trials were impressive on good days but still produced sudden dangerous mistakes, making it more stressful than helpful.
- A number of buyers of FSD feel misled about promised capabilities and timelines, describing the purchase as a waste or “party trick.”
- A smaller group reports long trips (hundreds to thousands of miles) mostly on FSD and considers it a “game changer,” insisting critics underestimate its current capability.
Legal responsibility & regulation
- Consensus that under current law the human driver receives any ticket and bears primary liability, since FSD is officially Level 2.
- Some argue the manufacturer should share or assume liability once the system is marketed as self‑driving; Mercedes’ limited‑scope Level 3 system is cited as an example of a company taking responsibility.
- Concerns are raised that owners pay for damage and higher insurance premiums caused by FSD mistakes, while Tesla bears little direct cost.
Comparisons with other systems
- Waymo is repeatedly contrasted as more conservative and statistically better documented, with some pedestrians and cyclists saying they feel safer around Waymo vehicles.
- Others note Waymo’s own failures (wrong‑way driving, hitting a pole, a bike crash) and question the degree of hidden remote human oversight.
- Disengagement data: Waymo reports to California DMV and shows far more miles per intervention than community‑reported Tesla FSD numbers, though environments and definitions differ.
Level 2 vs “Full Self‑Driving” branding
- Many emphasize that Tesla’s FSD is legally and practically SAE Level 2 and requires constant supervision.
- Several argue the “Full Self‑Driving” name and past promises (e.g., near‑term Level 5, cross‑country sleeping) are misleading or fraudulent.
- Others counter that, if treated as advanced driver assistance rather than autonomy, FSD can be a useful though imperfect aid.