Reuters Says Tesla Inflated FSD Safety Claims 3-Fold as 7 of 9 AI Trainers Distrusted It
Updated
Updated · Electrek · May 28
Reuters Says Tesla Inflated FSD Safety Claims 3-Fold as 7 of 9 AI Trainers Distrusted It
2 articles · Updated · Electrek · May 28
Summary
Reuters found Tesla’s “Full Self-Driving” safety claims were overstated by about threefold because the company compared its airbag-deployment crashes with broader federal tow-away crash data.
That mismatch cut Tesla’s touted “10 times safer” claim to roughly three times farther between comparable crashes, and researchers said even that was unreliable given Tesla’s 4.1-year fleet versus the U.S. average of 12.8 years.
Seven of nine former data labelers told Reuters they would not trust FSD to drive them, describing failures around emergency vehicles, construction zones, pedestrians and speeds 20 to 30 mph over limits.
Reuters also said Tesla heavily mapped robotaxi launch zones before deployments, with about 300 Utah labelers helping annotate Austin routes—undercutting Elon Musk’s claim that Tesla avoids laborious local mapping.
The findings land as NHTSA has four active FSD and Autopilot probes, while Tesla’s own website still says current features require active driver supervision and do not make vehicles autonomous.
The Reuters investigation, published in May 2026, uncovered major flaws in Tesla’s Full Self-Driving (FSD) technology by exposing misleading safety claims and widespread internal skepticism within the company. The report revealed that Tesla’s public safety metrics for FSD were both misleading and incomplete, as the company compared FSD’s performance to a broad national average of all human drivers instead of making fair comparisons under similar conditions. This approach obscured the true accident rates of FSD and failed to distinguish between different versions of the technology, raising serious concerns about the accuracy and transparency of Tesla’s safety claims.