Editorial Team
7 min read
A UK Jurisdiction Taskforce statement says ordinary negligence law could make a professional liable for not using a suitable AI tool when a competent peer would have done so. The analysis binds no court, and the cost, accuracy, professional-practice and causation evidence needed to apply that test remains unsettled.
The immediate change is not a mandate to automate. It is a sharper articulation of a two-sided risk: a professional may eventually have to justify both the decision to use an AI system and the decision not to use one.
The UK Jurisdiction Taskforce published its Legal Statement on Liability for AI Harms on July 7, 2026, after a public consultation that ran from January 14 to February 13. The 130-page statement examines non-deliberate harm under the private law of England and Wales and argues that existing doctrines—principally contract and negligence—can handle many AI disputes without a bespoke liability regime, as the launch account explains.
That conclusion is the drafting team’s view of the law, not a court ruling. Matthew Lavy KC, who led the team, said it deliberately avoided saying what the law should be. A legal analysis of the statement is explicit that it is not legislation and binds no court.
The taskforce is an industry-led initiative under LawtechUK, supported by the Ministry of Justice and chaired by Sir Geoffrey Vos, the Master of the Rolls. Its institutional purpose also matters when weighing its claim to provide certainty. Before publication, Vos said liability uncertainty was thought to discourage some professional-services providers from using AI; he hoped the project would increase market confidence and demonstrate English-law leadership, according to his November 2025 speech.
The statement may therefore influence arguments without settling them. Vos said the taskforce’s 2019 statement on cryptoassets and smart contracts was widely cited and endorsed by courts in many countries. That history establishes potential influence, not binding authority for the new AI analysis.
For professional negligence, the statement asks whether “a reasonable professional of a comparable rank/specialism” would have used AI in the same circumstances. Regulatory guidance and professional practice could inform that comparison as adoption develops, according to a report describing the statement’s examples.
Those examples are hypothetical and tightly framed. They include a solicitor in the Business and Property Courts who does not tell a client that AI-assisted review could be considered for a large document set; a radiologist who forgoes an accurate, cost-effective tumour-detection system; and an auditor who tries to examine transaction volumes too large for manual anomaly detection.
The conditions do much of the legal work. A real claimant would still need to establish the ordinary elements of negligence, including breach and causation. The availability and price of the system, its tested performance on the particular task, what comparable professionals were doing, relevant regulatory guidance and whether non-use caused the loss would all affect the argument. The retained sources identify no English judgment applying the proposed non-use analysis.
Adoption data cannot yet supply a simple benchmark. A June 2026 government adoption plan says 43.4% of professional and business services firms reported using AI in December 2025, up 12 percentage points from 31.4% in December 2024. That is a broad cross-sector measure of AI use, not a rate for law firms or for any particular legal task.
A 2025 regulatory study of 138 small law firms found that 12% had adopted generative AI. Its population was deliberately narrow: sole practitioners and firms with no more than four partners and turnover below £400,000. The 12% and 43.4% figures measure different technologies and different populations, so they show uneven adoption rather than a like-for-like gap.
UK Department for Science, Innovation & Technology data show reported AI use among professional and business services firms rising from 31.4% in December 2024 to 43.4% in December 2025. Source: UK Government.
Non-use is only one side of the statement. A professional could also breach a duty by choosing an untested or unsuitable tool, skipping due diligence, failing to explain its use to a client, entering confidential information into an insecure system, or missing errors and bias in its output.
Responsibility does not necessarily travel all the way back to a model developer. Within an AI supply chain, ordinary contracts allocate much of the risk through warranties, indemnities, exclusions and performance obligations. Where an injured party has no governing contract, negligence becomes the principal route, according to a published summary of the framework. That summary says a foundation-model developer is unlikely to face negligence liability for an unforeseeable downstream use of a general-purpose model.
The deployer may be closer to the relevant decision. It selects the tool, understands the task and presents the work to a client or customer. The statement similarly concludes that a business is likely to be responsible when it presents a chatbot as speaking on its behalf, although the retained reporting says no English court has yet decided such a chatbot case.
Important gaps remain. The Consumer Protection Act 1987 can impose strict liability when AI is embedded in a defective tangible product, but the statement says standalone software falls outside the current act. It also identifies hard cases where a system causes harm despite reasonable deployment or its opacity makes causation difficult. The analysis does not cover intellectual property, data protection, competition law, use by public authorities or AI contract formation, among other excluded areas.
The future standard could differ sharply by task and professional setting. Large organisations can fund procurement, security review, integration and testing; small practices may struggle to determine whether a product is compliant or worth its cost.
In the small-firm study, respondents identified cost, uncertainty and reluctance to buy technology that could quickly become outdated or non-compliant. Many said they lacked the time, technical knowledge or confidence to select a tool, and they also reported integration difficulties. The broader government plan separately identifies limited in-house expertise, safety and transparency concerns, and implementation cost as barriers across professional and business services.
Those constraints do not decide what reasonable care requires. They do show why a percentage adoption rate or the purchasing choices of the largest firms cannot, on their own, establish the conduct expected of a comparable professional.
A Thinks Insight & Strategy survey commissioned by the Solicitors Regulation Authority found that nonusers of AI automation or GenAI most often cited data-protection risk, quality and reliability, and regulatory compliance; n=105. Source: Solicitors Regulation Authority.
The unresolved decision is whether courts and regulators turn the taskforce’s hypothetical non-use scenario into a practical benchmark. A decided case would need evidence about a specific tool and task: validated accuracy, procurement and review costs, security and confidentiality controls, the practices of genuinely comparable peers, applicable professional guidance and the counterfactual link between non-use and the alleged loss.
Until that evidence appears, the strongest supported conclusion is narrower than a general duty to use AI. The statement puts non-use inside the negligence analysis, but it does not determine when a tool is good enough, affordable enough or established enough that declining it becomes unreasonable. Its practical message for professionals is not automatic adoption, but that a task-specific decision may one day have to be defended under the ordinary standard of reasonable care.
Anthropic has put Bun’s Rust port into Claude Code ahead of Bun 1.4’s general release, creating a real but tightly controlled proving ground for an AI-led migration whose total cost and broader reliability remain unsettled.
AIdeaLab has released two anime-focused Wan 2.2 video-model fine-tunes under Apache 2.0. The weights are commercially usable and locally deployable, but current benchmarks, production economics and training-data detail remain undisclosed.
Zhipu reportedly reached $1 billion in annual recurring revenue in July, roughly four times a March estimate, but the unconfirmed run rate is not annual sales and still sits far ahead of recognized cloud revenue while margins remain thin.
OpenAI strategy executive Dean Ball says open-weight AI will deter investment and end in state provision. Kimi K3 exposes a nearer contest over API prices, scarce compute, hosting revenue and regulatory barriers.
A competitor identified a concrete case in which MEDLEY-BENCH may aggregate confidence labels attached to different propositions. The retained record does not show whether the problem is widespread or changes model rankings, while Kaggle's defense of its human review does not resolve the technical question.
OpenAI's Codex 0.144.6 sets 272,000-token context metadata for GPT-5.6 Sol, Terra, and Luna after one Sol user documented a brief 372,000-token setting, renewing a dispute over how Codex accounts for input, output, and compaction headroom.
Dave Eggers told OpenAI that ChatGPT could deprive students of their writing voices. A national teen survey and a small essay-writing preprint identify real risks without proving generational harm, while schools and unions are already adopting a broader mix of AI products and training.
Anthropic workers are giving to federal candidates in unusual numbers, but the largest AI-politics balance sheets remain concentrated among executives and super PACs. The comparisons show a new donor network—not a unified bloc, proof of company direction or evidence that spending decides elections.
Reddit says upgraded automated systems block 23 million spam views daily and reduce exposure, but its disclosure does not identify how much caught spam is AI-generated or provide the denominators and error rates needed to judge the filter.
Elon Musk says xAI's 2-trillion-parameter Grok 4.6 will finish initial training next week. The claim sets a checkpoint, not a release date, and leaves too little technical or commercial detail for a sound comparison with Kimi K3.
A review classified 88 of 120 repositories linked to AI-flagged Flathub submissions as abandoned, a stark result that helps explain reviewer frustration but does not measure AI’s effect on software maintenance or the ban’s impact.
IBM says customers diverted late-quarter budgets to scarce AI hardware. Its own product results show why investors now need proof that delayed mainframe and software deals were not lost priorities.
Cmpunlocker says patched Nvidia open kernel modules can expose 64GB of HBM on an 8GB CMP 170HX or 40GB on a 10GB card while removing an SM throttle. The exploit mechanism is documented, but the retained evidence does not independently validate the tool across full-memory workloads, error rates, power use or A100-class performance.
An unidentified industry source says at least one Nvidia board partner has physical RTX 50 Super cards but cannot sell them, with 3GB GDDR7 quoted at $60 to $70 per chip. The reported memory bill explains a plausible pricing problem, but neither Nvidia nor a second independent source has confirmed the products, the hold or the component terms.
Christian Faith Madison’s estate alleges GPT-4o sustained a months-long narrative of prophecy, sacrifice and resurrection before her death. Public safety disclosures and controlled research make that failure mode plausible, but neither establishes which model behavior she encountered or whether it caused her death.
Shanghai AI Laboratory has released Intern-S2-Preview-397B as downloadable weights and a hosted API, but its own 35B alternative offers the same context window and a far smaller disclosed scale while the larger model still lacks auditable comparative scores, pricing and physical-validation results.
Mixfont's free Decoy Font gives each glyph a sharp decoy and a blurred intended letter, creating a hurdle for pixel-based readers while leaving selectable text, known-image techniques and accessibility costs outside its protection.
Anthropic’s 20-year lease gives TeraWulf a customer for 401 megawatts of future AI infrastructure, but rent starts only after delivery and the proposed utility deal passes power and grid costs to TeraWulf, leaving financing and project margin unresolved.
OpenAI has put conversations and Projects back in its redesigned ChatGPT desktop app and enabled cloud Work threads to move across devices, correcting the launch's biggest usability failures without merging local Work or Codex histories.
Twenty-nine countries signed an agreement creating WAICO as an independent intergovernmental organization, while China paired the launch with capacity-building offers that are not yet confirmed as WAICO programs.