Skip to main content
Open questionOpen question

Open question: who is liable when a behaviour the robot learned causes the harm?

A robot that acquired a behaviour after sale, from its own operation or from a fleet update, can injure someone without any component failing. The EU withdrew the directive written for that case, the revised product liability rules reach software without saying how learning is judged, and no other jurisdiction has answered it at all.

The verdict

Open question

No settled answer exists. The ledger poses the question, links the live debate, and does not answer it.

Key facts

What the sources say

Record ID
PAG-2026-0029
Kind
Open question
Jurisdiction
Global
Last verified
Added
  • The AI Liability Directive proposal, which would have eased proof for exactly this case, was withdrawn and published as withdrawn in the Official Journal on 6 October 2025.
  • Directive (EU) 2024/2853 makes software a product, so the learned component is inside the product, while leaving defectiveness to be judged case by case.
  • ISO 21448 gives a vocabulary for harm with no broken part, but it is a safety framework and allocates no liability.

What it changes

For a robot maker or a deployer

Nobody can tell you today how a court will treat a behaviour your robot taught itself, so the only defence you control is the record: what the model was at each release, what you tested, what the fleet reported and what you did about it. Build that record as a product feature rather than as a legal afterthought.

Sources

What this record was verified against

  1. EUR-Lex, Official Journal C/2025/5423, withdrawal of Commission proposalsPrimary · 6 October 2025
  2. EUR-Lex, Directive (EU) 2024/2853 on liability for defective productsPrimary · 18 November 2024

Related

Records that sit beside this one

Cite this record

Free to reuse under CC BY 4.0, with attribution. The record ID PAG-2026-0029 is permanent and is never reused.

In a sentence
According to the GAGE Physical AI Governance Atlas (as of 15 September 2026), open question: who is liable when a behaviour the robot learned causes the harm?.
APA
GAGE (Global Academy of Generative-AI Education). (2026). Open question: who is liable when a behaviour the robot learned causes the harm?. Physical AI Governance Atlas. Retrieved 15 September 2026, from https://www.gage.academy/tools/physical-ai-governance-atlas/records/PAG-2026-0029-question-liability-for-a-learned-behaviour
MLA
"Open question: who is liable when a behaviour the robot learned causes the harm?." Physical AI Governance Atlas, GAGE (Global Academy of Generative-AI Education), 15 September 2026, https://www.gage.academy/tools/physical-ai-governance-atlas/records/PAG-2026-0029-question-liability-for-a-learned-behaviour.
Chicago
GAGE (Global Academy of Generative-AI Education). "Open question: who is liable when a behaviour the robot learned causes the harm?." Physical AI Governance Atlas. Last modified 15 September 2026. https://www.gage.academy/tools/physical-ai-governance-atlas/records/PAG-2026-0029-question-liability-for-a-learned-behaviour.
Permalink
https://www.gage.academy/tools/physical-ai-governance-atlas/records/PAG-2026-0029-question-liability-for-a-learned-behaviour

Last updated . Every record re verified . The ledger is checked monthly, first Monday, and the same day for any new law or standard publication.

Back to the full ledger, or every record for Global and every open question record.

GAGE briefings tell you which AI regulation deadlines are coming, what they actually require of you, and when a program opens.

Powered by GAGE, free EU AI Act tools from the Global Academy of Generative-AI Education
CallEmailHelp