Disengagement statistics are theater; risk-factor language is the disclosure that has legal teeth. In its FY2025 10-K — surfaced through EdgarBeast and filed at sec.gov — Tesla writes that the regulatory patchwork increases the legal complexity with respect to autonomous vehicles. That single phrase reframes the whole robotaxi conversation.
A 'patchwork' is an admission that there is no single switch to flip. Some jurisdictions are permissive, some are considering rules, some are silent, and the boundaries do not line up with where customers or routes are. An AV company therefore cannot ship a national product the way it ships a phone app; it must clear approvals one legal map at a time, and each map can change under it.
This is the constraint that timelines tend to ignore. A demo proves the perception-and-planning stack can handle a route. It does not prove the company has the legal right to run that route commercially, without a safety driver, at scale. The 10-K's own words put the gating factor outside engineering and inside a fragmented approval process the company does not control.
It is also why the camera-versus-LiDAR debate is, partly, a regulatory bet. A vision-only stack and a mapped-LiDAR fleet face the same patchwork; neither sensor choice exempts a company from jurisdiction-by-jurisdiction permission. The honest read separates 'the system can drive here' from 'we are allowed to operate here commercially,' and the filing is candid that the second question is unresolved and fragmented.
When a robotaxi rollout slips, the press release blames 'an abundance of caution.' The sec.gov risk factor told you the structural reason in advance: a patchwork is slow to clear by design. Read the risk section before the keynote — it is the part the company is legally obligated to be honest in. Evidence indexed by EdgarBeast.