> ## Content Index
> Fetch the complete content index at: https://www.humanoidthreats.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Who Is Liable When a Hacked Robot Causes Harm?
- URL: https://www.humanoidthreats.com/who-is-liable-when-a-hacked-robot-causes-harm/
- Published: 2026-11-05T06:59:14.000Z
- Updated: 2026-11-05T06:59:14.000Z
- Description: Robot makers are usually liable first: EU rules from 9 Dec 2026 cover security flaws and missed patches. Learn how hacked robot liability is shared.
- Author: Noa Lindqvist
- Tags: Q&A, Regulation

In the EU, the robot's manufacturer is usually first in line when a hacked robot hurts someone, and from 9 December 2026 that covers software, security flaws and missing updates. A hacker's attack does not reduce the manufacturer's liability, though owners who skip updates or modify the robot can share the blame.

*Last updated: November 5, 2026\. This is general information, not legal advice. Liability depends on the country, the contract and the facts of each case.*

## Hacked robot liability in the EU: what changes on 9 December 2026

The EU's new Product Liability Directive, [Directive (EU) 2024/2853](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L%5F202402853&ref=humanoidthreats.com), "shall apply to products placed on the market or put into service after 9 December 2026" (Article 2(1)). It replaces the 1985 directive and makes three changes that matter directly for humanoid and other connected robots:

- **Software is a product.** The Directive's definition of product "includes ... software", and recital 13 says this holds "irrespective of the mode of its supply or usage". The control stack, the AI model and the companion app are all in scope.
- **A security flaw can make a robot defective.** Recital 32 says "A product can also be found to be defective on account of its cybersecurity vulnerability, for example where the product does not fulfil safety-relevant cybersecurity requirements."
- **Liability continues after sale.** A product stays "within the manufacturer's control" where the manufacturer can supply software updates or upgrades, itself or via a third party (recital 19). Recital 51 says manufacturers should not be exempt where harm results from "their failure to supply the software security updates or upgrades that are necessary to address those products' vulnerabilities".

The Directive does not itself require manufacturers to ship updates (recital 51 says so). It makes the cost of not shipping them land on the manufacturer when someone is harmed.

## Doesn't the hacker take the blame?

The attacker is liable for the attack under criminal and civil law, but attackers are rarely found and seldom able to pay. The Directive closes that gap. Recital 55 says the economic operator's liability "should not be reduced or disallowed" because of a third party's act, naming "a third party exploiting a cybersecurity vulnerability of a product" as the example. An injured person can claim against the manufacturer without first finding the hacker.

## Can the robot owner or operator be liable?

Yes, in two ways:

| Who                                                   | When they can carry liability                                                                         |
| ----------------------------------------------------- | ----------------------------------------------------------------------------------------------------- |
| Manufacturer (incl. software and AI component makers) | Defect at sale, a security vulnerability, or a missing security update within its control             |
| Anyone who substantially modifies the robot           | Treated as a manufacturer for the modified product (the Directive defines "substantial modification") |
| Importer or EU representative                         | When the manufacturer is outside the EU                                                               |
| Injured person or owner                               | Liability can be reduced where they "negligently failed to install updates or upgrades" (recital 55)  |
| Employer or deployer                                  | Under national workplace safety and negligence rules, separate from the Directive                     |

For a warehouse or factory, this means a deployer that disables auto-updates, adds its own control software or connects the robot to an unsegmented network may share the blame. Our [humanoid robot security checklist](https://www.humanoidthreats.com/humanoid-robot-security-checklist/) covers the controls that document reasonable care.

## Can a claimant prove a robot was hacked?

Proving a cyber defect is hard for an outsider. The Directive eases it: courts can order a defendant to disclose relevant evidence, and can presume defectiveness or causation where proof "would be excessively difficult" because of technical complexity (recitals 46 and 48). Contracts cannot waive this liability (recital 56). Damage covered includes death, personal injury including "medically recognised and medically certified damage to psychological health" (recital 21), property, and the destruction or corruption of non-professional data (recital 20).

## What other EU rules feed into liability?

The Directive looks at "safety-relevant cybersecurity requirements" when judging defects (recital 34). For robots, two such rule sets matter:

- **Cyber Resilience Act.** Since 11 September 2026, manufacturers must report actively exploited vulnerabilities and severe incidents, with an early warning within 24 hours and a notification within 72 hours. The main obligations apply from 11 December 2027 ([European Commission](https://commission.europa.eu/news-and-media/news/safer-and-more-secure-digital-products-2026-09-11%5Fen?ref=humanoidthreats.com)).
- **Machinery Regulation (EU) 2023/1230.** From 20 January 2027, its requirements on "protection against corruption" and on control systems that withstand reasonably foreseeable malicious attempts apply to machinery, including robots ([VicOne summary](https://lab-r7.vicone.com/blog/when-ai-moves-why-the-eu-machinery-regulation-matters-for-physical-ai-security/?ref=humanoidthreats.com)).

A manufacturer that misses these requirements gives a claimant a strong argument that the robot was defective. Our guide to [humanoid robot security standards](https://www.humanoidthreats.com/humanoid-robot-security-standards/) covers them in more detail.

## What about the United States?

There is no federal statute on hacked robot liability. Claims would generally run through state product liability and negligence law, under which courts ask whether a product was defectively designed, whether warnings were adequate and whether harm was foreseeable. How courts apply those tests to a cyberattack on a robot has not been settled by a reported robot case that we know of.

## Example: a real vulnerability, a hypothetical harm

The [Unitree G1 vulnerabilities](https://www.humanoidthreats.com/unitree-g1-vulnerabilities-cve-2026-76639-76640/) allowed root code execution, one over Bluetooth. Imagine (hypothetically) that a G1 sold in the EU after 9 December 2026 were taken over through a flaw like that and injured a worker. The worker could claim against the manufacturer or its EU importer, use the Directive's disclosure and presumption rules, and the attacker's role would not reduce the manufacturer's liability. If a patch had been released and the operator had refused to install it, the operator's share could reduce the claim. All known robot flaws are listed in the [Humanoid Robot Vulnerability Tracker](https://www.humanoidthreats.com/humanoid-robot-vulnerability-tracker/).

For how attackers get in, see our [humanoid robot attack surface map](https://www.humanoidthreats.com/humanoid-robot-attack-surface/) and [Can humanoid robots be hacked?](https://www.humanoidthreats.com/can-humanoid-robots-be-hacked/)

## Related questions

**Does the new EU liability law cover robots sold before 9 December 2026?** No. Recital 63 says it does not apply to products placed on the market or put into service before that date; the 1985 rules continue to apply to them.

**Is open-source robot software covered?** Free and open-source software developed or supplied outside a commercial activity is excluded (Article 2(2)). Open-source code built into a commercial robot is still part of that product.

**How long can someone wait before suing?** The Directive sets a general 10-year expiry period, extended to 25 years for personal injury with slowly emerging symptoms (recital 57).

## Sources

- [Directive (EU) 2024/2853 on liability for defective products, EUR-Lex](https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L%5F202402853&ref=humanoidthreats.com)
- [Reed Smith: EU Product Liability Directive and cybersecurity](https://www.reedsmith.com/en/perspectives/2025/07/eu-product-liability-directive-software-digital-products-cybersecurity?ref=humanoidthreats.com)
- [European Commission: Cyber Resilience Act reporting obligations](https://commission.europa.eu/news-and-media/news/safer-and-more-secure-digital-products-2026-09-11%5Fen?ref=humanoidthreats.com)
- [VicOne LAB R7: EU Machinery Regulation and physical AI security](https://lab-r7.vicone.com/blog/when-ai-moves-why-the-eu-machinery-regulation-matters-for-physical-ai-security/?ref=humanoidthreats.com)

Get the weekly Humanoid Threats Brief → [Subscribe](#/portal/signup)