SUMMARY: A Tesla technician’s $51 million lawsuit sits inside a much larger, harder-to-see problem: as robots have gained the mobility to work unguarded alongside people, the safety standards meant to cover them have stayed fragmented — voluntary in key places, or simply older than the robots now on the floor.
- A former Tesla technician is suing the company and robot maker Fanuc for $51 million after an industrial robotic arm allegedly struck him with roughly 8,000 pounds of force in 2023 — one of at least 77 robot-related incidents researchers identified in U.S. OSHA severe-injury data between 2015 and 2022
- OSHA has no robot-specific safety standard at all; it relies on a general workplace-safety clause dating to 1970, leaving the technical rules to a patchwork of industry standards
- The fastest-growing category — autonomous mobile robots (AMRs), which now operate unguarded alongside workers — is governed only by a voluntary standard, confirmed directly by the organization that wrote it
- Amazon’s own robotic-warehouse injury data is more complicated than either side’s talking points suggest: one rigorous academic study found severe injuries fell 40% in robotic facilities, while non-severe injuries rose 77%
8,000 Pounds, Without Warning
On July 22, 2023, robotics technician Peter Hinterdobler was helping a Tesla engineer disassemble an industrial robot at the company’s Fremont, California factory. According to a complaint filed in federal court, the robot’s arm “suddenly and without warning released with great force” — a release the filing compares to being struck by an 8,000-pound counterbalance weight. Hinterdobler alleges he was thrown to the floor and lost consciousness. He’s now suing Tesla and the robot’s manufacturer, Fanuc, for $51 million, citing roughly $1 million in medical expenses so far.
One detail in the complaint stands out: the robot had reportedly been relocated to an area of the facility “not designated for such equipment” shortly before the incident. If accurate, that points to a facility-management failure layered on top of whatever mechanical cause is ultimately established — not simply “a robot malfunctioned,” but a robot operating somewhere its safety setup hadn’t been designed for.
Tesla has moved to dismiss the case, arguing California’s workers’ compensation law bars the claim. Fanuc, Tesla, and Hinterdobler’s attorney did not respond to Manufacturing Dive’s requests for comment at the time of that outlet’s reporting.
One Lawsuit, Inside a Documented Pattern
Hinterdobler’s case is dramatic, but it isn’t an outlier in the way a single lawsuit might suggest. Researchers analyzing OSHA’s severe-injury reporting data identified 77 robot-related workplace incidents between 2015 and 2022, resulting in 93 injuries — finger amputations, and fractures to the head, torso, legs, and feet, among them. That figure almost certainly understates the true count. OSHA’s severe-injury program only captures incidents meeting a specific reporting threshold: hospitalizations, amputations, and loss of an eye. Injuries that don’t clear that bar — the kind more common with mobile robots than fixed industrial arms — wouldn’t show up in this dataset at all.
A separate, earlier case adds context rather than volume. In 2021, a KUKA-brand industrial robot at Tesla’s Giga Texas facility pinned an engineer against a surface, leaving him bleeding after its claws scratched his back and arm. The incident wasn’t publicly reported until The Information disclosed it in December 2023. Individually, these are anecdotes. Together with the OSHA figures, they describe a recurring category of workplace hazard that predates the current wave of AI-driven robotics investment by years.
The Amazon Data Is More Complicated Than the Headlines
No company’s robotic deployment has drawn more injury-rate scrutiny than Amazon’s, and the public numbers genuinely conflict. The Strategic Organizing Center’s 2022 analysis of Amazon’s own OSHA-reported 2021 data found a serious-injury rate of 7.3 per 100 workers at Amazon’s robotic sortable facilities, 28% higher than the 5.7 rate at its non-robotic sortable facilities. Amazon disputes comparisons like this one. A company spokesperson told Fortune that “claims that we’re significantly worse than others are false.” The spokesperson pointed to a 69% reduction since 2019 in the rate of injuries serious enough to require time away from work — a narrower metric than the overall recordable-incident rate critics typically cite.
The most methodologically serious independent study available complicates both sides of that argument rather than settling it. Researchers from Boston University, George Mason University, and IE University analyzed data from more than 140 Amazon fulfillment centers between 2016 and 2020. They found that robotic facilities saw a 40% decrease in severe injuries compared to non-robotic ones — but a 77% increase in non-severe injuries. That’s not simply “robots made things worse” or “robots made things better.” It’s a shift in the type and severity of harm, plausibly tied to algorithmic work-pacing rather than to physical contact with a robot at all. That distinction matters, because it points toward a different category of fix — work-rate design — than the one usually proposed, which is better robot guarding.
Why There’s No Single “Robot Safety Law”
Underneath both the Tesla lawsuit and the Amazon injury debate sits a structural fact: there is no U.S. federal regulation written specifically for workplace robots. OSHA has never issued a robot-specific standard; when it inspects a facility with robots, it applies the General Duty Clause, a broad requirement from the original 1970 OSH Act that employers keep workplaces free of recognized hazards. That’s a real enforcement tool, but it’s a catch-all, not a technical specification for how a robot should be guarded, sensed, or speed-limited.
The actual technical detail lives in a set of voluntary industry standards instead, developed by A3 (the Association for Advancing Automation) and accredited by ANSI, the American National Standards Institute. Fixed industrial robots — the kind that struck Hinterdobler — fall under ANSI/RIA R15.06, which requires a formal risk assessment and physical guarding. Collaborative robots, or “cobots,” fall under a revised pair of standards, ISO 10218-1 and ISO 10218-2, published in 2025 after nearly eight years of development, setting specific force and pressure limits for contact with a human. Both categories, in other words, have had their governing standards recently updated to reflect how these robots are actually used now.
The Standard That’s Still Voluntary
Autonomous mobile robots — machines that navigate a facility independently using onboard sensors rather than a fixed path — are the newest and fastest-growing category. They increasingly work in open floor space rather than behind a cage, and their governing standard hasn’t kept the same pace.
ANSI/RIA R15.08, first published in December 2020 and expanded with a second part in 2023, is the closest thing AMRs have to a dedicated safety standard. Carole Franklin, A3’s director of robotic standards development, was direct about its status in comments to industry outlet MūL Technologies: “The standards are voluntary industry consensus standards, so they’re not a regulation.” Franklin noted that compliance is nonetheless “highly recommended.” If OSHA inspects a facility after an incident and asks whether its mobile robots comply with R15.08, a company with no answer has effectively demonstrated it skipped its own due diligence — even though nothing legally required that diligence in the first place.
Amazon’s own fleet illustrates why this particular gap matters. Proteus, introduced in 2022, is described by the company as its first fully autonomous mobile robot built to move around employees without the caged separation Amazon’s earlier robots required — exactly the category R15.08 was written to cover. It’s also exactly the category still governed only by a standard its own authors call voluntary rather than mandatory.
What Comes Next Doesn’t Look Like It’s Slowing Down
The regulatory gap around AMRs isn’t a closed problem; if anything, it’s about to get a harder version of itself. Humanoid robots currently operate under the same general standards as other collaborative machines. Tesla’s Optimus program is the most aggressively scaled example, though its production numbers are contested territory. Tesla reportedly delivered only hundreds of units in 2025 against an earlier target of roughly 5,000, and Musk himself said in early 2026 that current units are “primarily for learning, not productive tasks.” Every major Optimus production target since 2022 has reportedly been missed, according to one industry tracker — worth keeping in mind whenever a specific unit-count for 2026 or beyond is cited as settled fact, including elsewhere in this account.
A humanoid-specific safety standard does not yet exist. ISO 25785-1 remains at the working-draft stage within ISO’s Technical Committee 299, with no published completion date as of mid-2026. Until it arrives, a worker injured by a humanoid robot has product-liability law to fall back on, but no robot-specific regulatory violation to point to. It’s the same structural gap AMRs have occupied since 2020, arriving in an even less-tested form.
The Parts of This Story That Aren’t Settled Yet
Several things here resist a clean conclusion. Whether Amazon’s injury-rate figures are genuinely worse than industry peers, or simply more visible because of its scale and the intensity of outside scrutiny it draws, isn’t something the public data cleanly resolves. The comparisons cited above use different states, different years, and different methodologies, and Amazon disputes the framing of several of them. The GMU severity-shift finding is the most rigorous single data point available, but it describes Amazon’s facilities specifically; whether the same severe-down/non-severe-up pattern holds at other robotic warehouse operators isn’t established by any source reviewed here.
It’s also not possible to say, from public information, how many incidents involving AMRs specifically — as opposed to fixed industrial arms — are undercounted in OSHA’s severe-injury data, given that program’s reporting thresholds. And R15.08’s own Part 3, covering requirements for the companies that deploy IMRs rather than the manufacturers that build them, was still being developed as of the most recent sourcing available here. Even the voluntary framework that exists, in other words, isn’t yet complete on the deployment side.
One scope note: this account focuses on physical-safety and regulatory-standard questions around industrial and mobile robots. It doesn’t cover the separate, related question of robot and industrial-control-system cybersecurity — a genuinely adjacent risk category, and one worth its own treatment rather than a passing mention here.
Where This Leaves the Argument
Robots didn’t get more dangerous by design between 2020 and now. They got more mobile, more autonomous, and more willing to share open floor space with people who used to be separated from them by a cage. The safety architecture built for the caged version hasn’t fully caught up to the uncaged one. Cobots got an updated standard in 2025. Fixed industrial robots have had a mature one for years. AMRs, the category actually driving the current wave of warehouse and factory investment, are still governed by a standard its own authors describe as advice rather than law.
That gap won’t close itself, and it isn’t obviously anyone’s fault in particular — voluntary consensus standards take years to build and even longer to make mandatory, if they ever do. A technician thrown to the floor by 8,000 pounds of uncontrolled force, and a warehouse worker hurt not by a robot’s arm but by the pace a robot sets, are different injuries with different causes. Both, in their own way, are standing in the space between how fast this technology moved and how fast its rules did.
