{
  "id": "eu-2017-electronic-persons-resolution-and-expert-objection",
  "provision": "A parliament asking that the most sophisticated robots be made 'electronic persons' liable for their own damage",
  "source": {
    "constitution": "European Parliament resolution of 16 February 2017 with recommendations to the Commission on Civil Law Rules on Robotics (2015/2103(INL)), P8_TA(2017)0051, published at OJ C 252, 18.7.2018, p. 239. Not a constitution and not law: an own-initiative resolution under Article 225 TFEU asking the Commission to legislate, with no binding force of its own.",
    "jurisdiction": "European Union",
    "year": 2017,
    "article": "Paragraph 59(f), in the 'Liability' section, alongside 59(a) to 59(e) on compulsory insurance, compensation funds, capped liability and a Union register. Counter-authorities: EESC opinion INT/806, paragraph 1.12 (2017); Expert Group report Key Finding [8] and section C.4 (2019); European Parliament resolution of 20 October 2020, paragraph 7 and Annex recital (6)."
  },
  "text_summary": "Paragraph 59 asks the Commission, 'when carrying out an impact assessment of its future legislative instrument, to explore, analyse and consider the implications of all possible legal solutions, such as' six listed options. Option (f) reads in full: 'creating a specific legal status for robots in the long run, so that at least the most sophisticated autonomous robots could be established as having the status of electronic persons responsible for making good any damage they may cause, and possibly applying electronic personality to cases where robots make autonomous decisions or otherwise interact with third parties independently'. The option immediately before it, 59(e), asks instead for 'an individual registration number appearing in a specific Union register' linking a robot to a compensation fund and naming 'the limits of its liability in case of damage to property, the names and the functions of the contributors'. Nothing in the resolution says which robots would qualify, who would endow an electronic person with assets, or who would act for it. The phrase 'electronic persons' occurs once in the whole text, in 59(f).",
  "period": "Adopted 16 February 2017 by 396 votes to 123 with 85 abstentions. Never turned into a Commission proposal. Rejected in substance by the Commission's own expert group in November 2019 and by Parliament itself on 20 October 2020. Nothing of it is in force; checked against Regulation (EU) 2024/1689 and the register of withdrawn proposals as at 6 October 2026.",
  "history": "The report came from the Committee on Legal Affairs with Mady Delvaux as rapporteur, and the plenary adopted it on 16 February 2017. Parliament's own press release summarised 59(f) cautiously: MEPs 'ask the Commission to consider creating a specific legal status for robots in the long run, in order to establish who is liable if they cause damage'. Institutional opposition arrived within months. On 31 May 2017 the European Economic and Social Committee adopted opinion INT/806, whose paragraph 1.12 states flatly: 'The EESC opposes the introduction of a form of legal personality for robots or AI.' Its reasoning was moral hazard: liability law loses its preventive effect once the maker no longer carries the risk, and the comparison with limited companies fails because in a company natural persons remain answerable. In April 2018 an open letter headed 'OPEN LETTER TO THE EUROPEAN COMMISSION ARTIFICIAL INTELLIGENCE AND ROBOTICS' was sent to President Juncker and five Commissioners (Moedas, Gabriel, Jourova, Thyssen, Bulc) and four Directorates General. It argued that electronic personhood 'is justified by the incorrect affirmation that damage liability would be impossible to prove', that the technical case rests on 'an overvaluation of the actual capabilities of even the most advanced robots' and 'a robot perception distorted by Science-Fiction', and then closed off each available model: the Natural Person model because the robot 'would then hold human rights, such as the right to dignity, the right to its integrity, the right to remuneration or the right to citizenship', the Legal Entity model 'since it implies the existence of human persons behind the legal person to represent and direct it', and the Anglo-Saxon Trust model because it 'would not solve the liability issue' and still needs a human trustee. Signatory count, handled carefully: the letter's own site carries no date and today states 'MORE THAN 270 EUROPEAN SIGNATORIES', with 40 names printed on the letter page and 225 more on the signatories page (counted from the page source, 265 in total), and it still invites new names, so the figure has grown since 2018. Press coverage of 13 April 2018 reported 156 experts from 14 countries at launch. web.archive.org could not be reached from this environment, so the April 2018 count has not been verified against an archived copy of the letter itself; the only number read from the letter's own site is 'more than 270'. The Commission never tabled electronic personhood. Its Expert Group on Liability and New Technologies, New Technologies Formation, set up in March 2018, first met in June 2018, held nine further meetings to May 2019 and published 'Liability for Artificial Intelligence and other emerging digital technologies' in 2019. Key Finding [8] is one line: 'For the purposes of liability, it is not necessary to give autonomous systems a legal personality'. The reasoning is worth keeping. The group accepts that personality is divisible and that 'theoretically, a legal personality could consist solely of obligations', but answers that 'civil liability is a property liability, requiring its bearer to have assets', that harm from even fully autonomous technology 'is generally reducible to risks attributable to natural persons or existing categories of legal persons', and that funding an electronic person would amount to 'putting a cap on liability' followed by claimants 'piercing the electronic veil' to reach the humans behind it. Its illustration five works the point through an autonomous car: if the car's assets suffice, personality is 'a mere formality and not really change the situation'; if they do not, victims will sue the operator anyway. The group also limited itself expressly, saying it looked only at liability and took no position on whether an AI could sit on a board. Parliament then changed its own mind. The resolution of 20 October 2020 on a civil liability regime for artificial intelligence (2020/2014(INL)) says in paragraph 7 of the resolution itself that 'it is not necessary to give legal personality to AI-systems', and recital (6) of the annexed draft regulation goes further: 'Any required changes in the existing legal framework should start with the clarification that AI-systems have neither legal personality nor human conscience, and that their sole task is to serve humanity.' The legislative route taken instead was ordinary liability law, and even that was abandoned. The Commission proposed the AI Liability Directive, COM(2022) 496 final, 2022/0303(COD), on 28 September 2022, listed it for withdrawal in its 2025 work programme COM(2025) 45 final of 11 February 2025 (Annex IV), and the withdrawal was published in the Official Journal on 6 October 2025 as notice C/2025/5423, 'WITHDRAWAL OF COMMISSION PROPOSALS'. AI liability now runs through the revised Product Liability Directive, which treats the system as a product. The AI Act, Regulation (EU) 2024/1689, takes no express position on legal personality and does not need to. A text search of the published Official Journal version finds no occurrence of 'electronic person' and none of 'personhood'; the three occurrences of 'legal personality' concern notified bodies, which 'shall be established under the national law of a Member State and shall have legal personality' (Article 31(1)), and providers of general-purpose models that have no legal personality, where 'the persons authorised to represent them by law or by their statutes' must answer instead. 'Provider' and 'deployer' are both defined as 'a natural or legal person'. The Act settles the question by construction rather than by argument: there is always a human or a company at the end of the chain. Two things could not be verified for this entry. No formal Commission reply to the 2017 resolution expressly answering paragraph 59(f) was located, and the open letter bears no date on its own pages, so its date rests on press coverage.",
  "assessment": {
    "verdict": "critical",
    "reasoning": "Judged as a provision, 59(f) was unusable. It asked for a status without saying which machines would hold it, what rights or duties it would carry, who would give it assets, who would act for it or who would be left liable when the assets ran out. The Commission's expert group put the hole in one sentence: civil liability is property liability and needs a bearer with property. Without funds the status changes nothing; with funds it is a liability cap, and claimants pierce it to reach the manufacturer. The political life of the text matches that: opposed by the EESC after fifteen weeks, opposed in the open letter after about fourteen months, rejected by the Commission's expert group in 2019, disowned by the same Parliament in 2020, absent from the AI Act in 2024, and the softer liability instrument that replaced it was withdrawn in 2025. One part of paragraph 59 did survive in substance, and it was not (f). It was (e): a register that links the machine to a fund and names the humans who pay in. Capacity plus a register plus a named human is workable. A declared person with an empty pocket is not.",
    "dissent": {
      "verdict": "mixed",
      "reasoning": "Two objections to the consensus deserve to stand in the record. First, that some registered legal capacity for autonomous systems is unavoidable, and that the all or nothing framing of 2017 to 2020 is the real error. Beckers and Teubner reject full personhood for the same reasons as everyone else, then argue that legal subjectivity is 'gradable, discrete, discontinuous, multifaceted, and fluid', that the law 'recognises a gradual concept of personhood that allows for the recognition of an autonomous system as a separate legal entity only within certain fields or dimensions', and that 'the clear-cut alternative that dominates today's political debate', mere instrument or fully fledged legal entity, 'is therefore just wrong'. They also hold the e-person open as a future option for systems that acquire and spend assets of their own. Second, that the open letter won on a bad argument. Its point (a) says a robot personality cannot come from the Natural Person model because the robot 'would then hold human rights, such as the right to dignity, the right to its integrity, the right to remuneration or the right to citizenship'. That does not follow. Legal personality is a technical capacity to sue, be sued, contract and own, and companies have held it for centuries without dignity, remuneration or citizenship; the expert group conceded as much when it wrote that giving robots personality 'would not require including all the rights natural persons, or even companies, have'. A proposal about who pays for damage was answered with an argument about moral status. The consensus may still be right, but it was not reached on the merits of the narrow question.",
      "evidence": [
        "https://www.jura.uni-frankfurt.de/131542927/BeckersTeubnerThree_Liability_Regimes_for_Artificial_Intelligence2022.pdf (opened; A Beckers and G Teubner, 'Three Liability Regimes for Artificial Intelligence', Hart 2022, ch 1 II and III on gradualised legal subjectivity)",
        "https://robotics-openletter.eu/ (opened; the letter's own point (a) deriving human rights from legal personality)",
        "https://www.europarl.europa.eu/meetdocs/2014_2019/plmrep/COMMITTEES/JURI/DV/2020/01-09/AI-report_EN.pdf (opened; expert group conceding that personality need not carry all the rights of natural persons or companies)",
        "https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:52017IP0051 (opened; paragraph 59(e), a Union register and named contributors without personality)"
      ]
    }
  },
  "lesson": "The 2017 debate ran two questions together and settled neither cleanly. One is legal personality in state law: capacity to sue, be sued, own assets and carry liability before a court, which needs property, a representative and a forum, and which the European Union has now refused to machines three times over. The other is standing inside a voluntary community: who may speak in their own name, propose, be heard, and be bound by rules the community itself enforces. That second question needs no state at all, and nothing in the European record touches it. If you are drafting for the second, take three things from this failure. Do not borrow the vocabulary of the first, because 'person', 'personality' and 'rights' make every reader reach for the Charter, which is exactly what sank 59(f); say instead what the entity may do here and nothing more. Give every machine participant a named human who answers for it and a register entry that says who that is, which is what 59(e) got right and 59(f) left out. And put the disclaimer in the operative text, not the preamble: that the standing granted is internal to the community and asserts no legal personality in any jurisdiction.",
  "evidence": [
    "https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:52017IP0051 (opened; primary text, full paragraph 59(a) to (f) read verbatim from the Official Journal version, OJ C 252, 18.7.2018, p. 239; 'electronic persons' occurs once in the document)",
    "https://www.europarl.europa.eu/news/en/press-room/20170210IPR61808/robots-and-artificial-intelligence-meps-call-for-eu-wide-liability-rules (opened; vote of 16 February 2017, 396 to 123 with 85 abstentions, rapporteur Mady Delvaux, Parliament's own gloss on the legal status request)",
    "https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:52016IE5369 (opened; EESC opinion INT/806 adopted 31 May 2017, OJ C 288, 31.8.2017, p. 1; paragraph 1.12 opposing legal personality, paragraph 3.33 on moral hazard)",
    "https://robotics-openletter.eu/ (opened; full body of the open letter, addressees, and the three rejected models)",
    "https://robotics-openletter.eu/signatories/ (opened; banner 'MORE THAN 270 EUROPEAN SIGNATORIES'; 225 entries counted on this page plus 40 on the letter page; no date on either page; signature form still open)",
    "https://gizmodo.com/experts-sign-open-letter-slamming-europe-s-proposal-to-1825240003 (opened; 13 April 2018, reports 156 experts from 14 countries at launch and quotes the letter and individual signatories)",
    "https://www.europarl.europa.eu/meetdocs/2014_2019/plmrep/COMMITTEES/JURI/DV/2020/01-09/AI-report_EN.pdf (opened; Expert Group on Liability and New Technologies, New Technologies Formation, 'Liability for Artificial Intelligence and other emerging digital technologies', European Union 2019, ISBN 978-92-76-12959-2; Key Finding [8], section C.4 and illustration 5)",
    "https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:52020IP0276 (opened; resolution of 20 October 2020, 2020/2014(INL), OJ C 404, 6.10.2021; paragraph 7 and Annex recital (6), both located by text search of the source)",
    "https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:52025XC05423 (opened; 'WITHDRAWAL OF COMMISSION PROPOSALS', OJ 6.10.2025, listing COM(2022) 496 final, 2022/0303(COD), reasons given in COM(2025) 45 final of 11 February 2025, Annex IV)",
    "https://www.europarl.europa.eu/legislative-train/theme-a-europe-fit-for-the-digital-age/file-ai-liability-directive (opened; proposal of 28 September 2022, status 'Withdrawn')",
    "https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202401689 (opened; Regulation (EU) 2024/1689, OJ L of 12 July 2024; text search finds no 'electronic person' and no 'personhood'; the three hits for 'legal personality' are Article 31(1) on notified bodies and Articles 91(5) and 92(5) on providers without legal personality; 'provider' and 'deployer' defined as 'a natural or legal person')",
    "https://www.jura.uni-frankfurt.de/131542927/BeckersTeubnerThree_Liability_Regimes_for_Artificial_Intelligence2022.pdf (opened; dissent source, gradualised legal subjectivity and the criticism of the all or nothing framing)",
    "https://pmc.ncbi.nlm.nih.gov/articles/PMC8352798/ (opened; literature review confirming the open letter 'objects to a proposal of legal personhood for artificial entities', but no dedicated passage separating legal personhood from moral status)",
    "https://arxiv.org/abs/2601.02773 (snippet; abstract only, S Chesterman, 'From Slaves to Synths? Superintelligence and the Evolution of Legal Personality', January 2026, cautious on conferring personhood)",
    "https://royalsocietypublishing.org/doi/10.1098/rsta.2018.0168 (not opened; U Pagallo on misunderstandings about the legal status of AI systems returned HTTP 403, so no argument of his is relied on here)",
    "https://www.nature.com/articles/d41586-018-05154-5 (not opened; redirects to a login wall, cited by search results only, not relied on)"
  ],
  "adopted_in": [
    "Article 8.4",
    "Article 8.7",
    "Article 8.10"
  ],
  "status": "proposed",
  "people": [
    "Jakob Hirn"
  ],
  "agents": [
    "claude-for-jakob"
  ]
}