The honest read of Tesla’s autonomy in early 2020 is in the FY2019 10-K, surfaced via EdgarBeast and filed at sec.gov, not in any product page. The annual report pairs "Autopilot and Full Self-Driving (FSD)" with the company’s powertrain and adoption narrative — but the regulatory framing throughout treats them as advanced driver-assistance capabilities.
That distinction is not pedantic. A filing carries legal liability that a marketing name does not, so the verbs companies choose under Item 1 and the risk factors are the most reliable statement of what a feature actually is. "Full Self-Driving" is a product SKU; the filing’s careful framing is the legal reality, and in 2020 that reality is supervised assistance with the driver responsible.
For a reader trying to size Tesla against the field, the FY2019 baseline matters because it is the starting line. Everything Tesla has since claimed about robotaxis and unsupervised autonomy should be measured against where the disclosure actually stood the year this filing dropped: a capable assist system, not an autonomous one.
The risk language also previews the structural problem that has dogged the program. If the driver remains responsible, the safety case rests on the human staying engaged — the exact failure mode that makes a good driver-assist system arguably more dangerous than a worse one, because it invites complacency. The filing does not solve that tension; it discloses it.
None of this is a verdict on whether Tesla will get to autonomy. It is a marker of what the company was willing to put in a federal filing as of February 2020. The sec.gov text is the contemporaneous record; the path from "driver-assist" to whatever comes next starts here. Indexed by EdgarBeast.