AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 495

Mr Laurence White

Submitter group: Individuals.

Position labels are not shown for individual submitters.

An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.

Open in the interactive guide

Summary

Read the original submission for the full text.

Argues that existing tort law already provides much of the machinery to allocate responsibility for AI-caused harm, but uncertainty remains about data damage, strict liability and frontier developers' duties. Recommends an ALRC reference on tortious liability, a mandatory safety-incident disclosure regime for frontier models, and international coordination.

Key points

Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.

  1. Argues that much of the legal machinery for allocating responsibility for AI harm already exists in tort law, and that the difficulty is uncertainty about whether it applies.

    Much of the legal machinery needed to allocate responsibility for harm caused by AI systems already exists in the law of tort. The difficulty is less an absence of doctrine than uncertainty about whether it applies.

  2. Argues that Australian courts already have the materials to hold developers or deployers liable in negligence and to set a demanding standard of care for frontier activities.

    Australian courts already have the materials to hold a developer or deployer liable in negligence for harm caused by an autonomous system and, applying Burnie, to set a demanding standard of care for frontier activities.

  3. Argues for strict liability for frontier systems with documented dangerous propensities whose residual risk reasonable care cannot remove.

    there is a respectable economic case for strict liability for the narrow class of frontier systems with a documented dangerous propensity whose residual risk cannot be removed by reasonable care: strict liability affects the level of an activity, not only the care taken in it, which matters when an activity is new and its dangers not well understood.

  4. Says it is not settled whether loss of or damage to data is damage to property, with the nearest authorities pointing in different directions.

    Whether loss of or damage to data is damage to property is not settled. The nearest authorities point in different directions.

  5. Argues that developers' terms may negate responsibility to users and, arguably, affected persons, subject to non-excludable warranties and unfair-contract-terms rules.

    First, frontier developers publish terms of use disclaiming responsibility in wide terms; on Mallonland, those terms may (subject to the non-excludable warranties and also to the unfair contract terms regime in the Australian Consumer Law)21 negate any assumption of responsibility to users and, arguably, to persons affected by them.

  6. Says Australia could condition the supply of frontier models on independent evaluation and mandatory incident disclosure to the AI Safety Institute.

    Australia has no frontier developer but many deployers, and could condition the supply of frontier models into the Australian market on the developer’s submission to independent evaluation and mandatory disclosure of safety incidents, here or in a recognised partner jurisdiction, with the Australian AI Safety Institute as recipient.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Consider Commonwealth, State and Territory legislation treating damage, corruption, loss or loss of control of data or computer systems as property damage for negligence and ACL Part 3-5, with consequential economic loss recoverable. Leave general pure-economic-loss law untouched.

    consideration be given to the Commonwealth, with the States and Territories, legislating to provide that, for the purposes of negligence and of Part 3-5, damage to, corruption of, loss of, or loss of control over data or a computer system is damage to property, with consequential economic loss recoverable on the same footing as for tangible property. This leaves the general law of pure economic loss untouched while removing the classification problem that would otherwise defeat the most likely class of claims.

  2. Align Australian AI liability and safety rules with those of the UK, EU and potentially US states, with mutual recognition of evaluator findings and incident reports among allied AI safety institutes.

    Australia is better placed to influence outcomes within a group of like-minded jurisdictions than alone. Its rules should be aligned with those of the UK and the EU and, potentially, of US states so that developers face a consistent standard, with mutual recognition of evaluator findings and incident reports among allied AI safety institutes.

  3. The Commonwealth Attorney-General, alone or with the Standing Council of Attorneys-General, should refer whether tortious-liability law needs reform to the Australian Law Reform Commission.

    The Commonwealth Attorney-General, alone or acting with the Standing Council of Attorneys-General (SCAG), should refer to the Australian Law Reform Commission the broad issue of 28 Your Response Ltd v Datateam Business Media Ltd [2014] EWCA Civ 281; [2015] 1 QB 41. 29 Property (Digital Assets etc) Act 2025 (UK) s 1, in force 2 December 2025. 30 UKJT Statement at [72]–[76] (software not embedded in a tangible thing is not “goods” under the Consumer Protection Act 1987 (UK)). 31 Amodei, “We must pace the frontier”, section “Embedded evaluators”. 32 Amodei, “We must pace the frontier”, sections “Pacing within democracies” and “Global pacing”. 5 whether law reform is needed to the law of tortious liability

  4. Consider tort-law reform for AI issues, including deeming damage, corruption, loss or loss of control of data or commercial computer systems recoverable property damage for ACL Part 3-5, negligence, strict liability and criminal damage.

    whether law reform is needed to the law of tortious liability so as to address novel issues raised by AI systems, including whether law reform would be desirable to provide that: a) damage to, corruption of, loss of, or loss of control over data or a commercial computer system is deemed to be damage to property and thus recoverable direct damage for the purposes of Part 3-5 of the Australian Consumer Law, the law of negligence and strict liability, and the law of criminal damage;

  5. As part of that reference, consider providing that a developer or deployer who knows or ought to know of an AI system's dangerous propensity is strictly liable for harm caused by its acts, subject to the usual tort defences.

    a developer or deployer who knows or ought to know of a dangerous propensity of an AI system is strictly liable for harm caused by acts of the AI system (subject to the usual defences in tort including novus actus interveniens, i.e. the intervening voluntary act of a third party); and

  6. As part of that reference, consider providing that a developer or deployer owes a duty of care to take reasonable care to prevent harm to persons foreseeably endangered, despite the generality of uses of the system.

    a developer or deployer of an AI system owes a duty of care to take reasonable care to prevent harm to persons foreseeably endangered by the AI system, notwithstanding the generality of uses to which the system may be deployed.

  7. Establish a mandatory safety-incident disclosure regime for frontier models supplied in Australia, administered by the AI Safety Institute, with market access conditioned on independent embedded evaluation here or in a jurisdiction recognised on the basis of equivalent regulation and supervision.

    Establish a mandatory safety-incident disclosure regime for frontier models supplied in Australia, administered by the Australian AI Safety Institute, with market access conditioned on independent embedded evaluation here or in a jurisdiction recognised on the basis of equivalent regulation and supervision.

  8. Step up coordination on AI safety with like-minded jurisdictions, including through the G20 and United Nations.

    Step up coordination of AI safety issues with like-minded jurisdictions, including through bodies such as the G20 and the United Nations.

Read the original

Open in the interactive guide