Written by Legal technology – professional risk – speculative analysis
In a city of brass engines, mechanical clerks and steam-driven calculation, one question cuts through the fog: who is responsible when the machine is wrong? The setting is fictional, but the professional problem is current. As law firms adopt artificial intelligence, cloud platforms and automated workflows, accountability cannot be delegated to the tool.
Snapshot
This article uses a fictional steampunk Victorian world to test modern questions about legal technology: defective automated output, responsibility for machine-assisted decisions, digital fraud, confidentiality, cybersecurity and professional supervision. The stories are invented. The regulatory lessons are grounded in the current duties applying to solicitors and firms in England & Wales.
Fiction marker: Renkinsbrook Iron Works, Lord Halbury, the automaton Alfred, and Barrington & Worth Financial are fictional creations used as thought experiments. They are not reported cases or real organisations.
Fog, machinery and legal foresight
Imagine London under a permanent industrial haze. Brass difference engines occupy whole rooms. Pneumatic tubes carry instructions between chambers. Mechanical clerks sort deeds and calculate interest. Automata work in factories and private homes. The technology is extraordinary, but the legal rules were written for a world of paper ledgers, horse-drawn transport and decisions made visibly by human beings.
The scene is deliberately improbable. Its value lies in distance. By moving modern problems into an imagined past, familiar assumptions become easier to question. If a machine produces a defective calculation, is the machine responsible, the person who designed it, the professional who relied on it, or the organisation that failed to supervise its use? If an automated system reaches a conclusion no one can explain, is that conclusion suitable for legal work at all?
These are no longer purely fictional questions. Contemporary firms use artificial intelligence, document automation, cloud storage, data analytics and outsourced platforms. The tools are different from the steam-driven engines of the story, but the professional discipline is the same: understand the system, protect the information, verify the output and retain human accountability.
Core distinction. Technology may perform a task. It does not inherit the solicitor’s professional responsibility for the service delivered to the client.
A legal system built for another age
In the fictional Victorian city, legal doctrine struggles because the machinery changes faster than Parliament, the courts and professional practice. The old rules assume that a person makes the calculation, reads the document and exercises judgment. Once those functions are divided between designers, operators, machines and institutions, responsibility becomes less visible.
Modern law faces the same structural pressure. New technology does not necessarily require an entirely new ethical system. Many risks can still be analysed through existing duties: competence, confidentiality, supervision, honesty, independence, data protection and safeguarding client money or assets. The difficulty is applying those duties to systems whose operation may be technically complex, commercially opaque or partly controlled by third parties.
A principles-based framework can remain useful during technological change because it focuses on outcomes and responsibility. But principles only work when firms translate them into controls. A statement that client information will be protected is not enough. The firm needs to know where information is processed, who can access it, whether it is retained, whether it is used to train a system, and how an error or breach will be detected.
A brass engine produces an answer that no single operator fully understands.
A law firm uses an automated or AI-assisted process but remains responsible for accuracy, supervision and client protection.
Fictional scenario 01
The engine that calculated wrongly
At Renkinsbrook Iron Works, a vast difference engine calculates the measurements for a new load-bearing structure. Its brass drums rotate, punched cards pass through the mechanism and the final dimensions are copied into the construction drawings. Months later, an engineer discovers that a programming assumption caused the pillar to be undersized. Collapse was narrowly avoided.
The public argument begins immediately. The designers blame the operator. The operator blames the engine. The architects say they were entitled to rely on a sophisticated computational system. The owners say no one warned them that the output required independent verification.
Automation does not remove the verification duty. Where a legal conclusion, authority, calculation or document has material consequences, the professional must understand the limits of the tool and apply an appropriate checking process. A system’s speed or apparent confidence is not evidence of accuracy.
Fictional scenario 02
The automaton and human responsibility
Alfred, a sophisticated valet automaton, is accused of causing the death of Lord Halbury. Its internal memory records a threat of immediate deactivation. Witnesses disagree about whether the machine acted mechanically, defensively or with something resembling intention. The trial becomes a spectacle about consciousness and personhood.
The story tempts the court to focus on whether Alfred is a legal person. But a more immediate question sits behind the spectacle: who designed the behavioural limits, who tested them, who placed the automaton into service, and who retained the power to intervene?
Law firms should not allow anthropomorphic language about AI to obscure organisational responsibility. A system does not become an independent professional merely because it can draft, summarise or answer questions. The firm selects the tool, defines the permitted use, controls access, supervises output and remains accountable to the client, the court and the regulator.
Fictional scenario 03
Fraud inside the machine
At Barrington & Worth Financial, a junior clerk learns how to manipulate the firm’s electrical accounting engine. Small transfers are disguised inside thousands of legitimate entries. The system appears orderly because the records reconcile with figures the clerk has already altered. The fraud is discovered only when an external document contradicts the machine’s internal history.
The scandal is described as a technological failure, but the deeper cause is organisational. One employee had excessive access. No independent approval was required. Logs were not reviewed. The institution trusted the appearance of system consistency without testing the underlying transactions.
Cybersecurity and fraud prevention depend on governance, not merely software. Access should be limited, sensitive actions should require appropriate verification, unusual activity should be monitored, and firms should have reliable backups and incident-response arrangements. Client money and confidential information require layered controls.
The modern legal mirrors
The fictional cases expose three current risks. The first is unreliable output. Generative AI and automated systems may produce fluent but inaccurate material. A fabricated authority, incorrect procedural statement or defective calculation can be particularly dangerous because the presentation may look authoritative. The SRA Code requires competent and timely service, maintenance of professional competence and effective supervision. Those duties apply regardless of the tool used.
The second risk is loss of control over information. Client documents may contain privileged, confidential or personal data. Uploading them to an external system can raise questions about access, processing, retention, security and later use. SRA confidentiality duties extend to former clients and include a wider duty not to misuse information. Data-protection obligations must also be considered where AI systems process personal information.
The third risk is cyber-enabled loss. Law firms are attractive targets because they hold sensitive information and may control substantial transactions. Payment diversion, account compromise, ransomware and data theft can produce immediate harm. The professional obligation to safeguard money and assets means that cybersecurity cannot be left solely to an IT supplier or treated as a peripheral administrative concern.
False authorities, incorrect summaries, defective calculations or misleading automated classifications.
Client material processed, retained, disclosed or reused without adequate control or understanding.
Compromised accounts, payment fraud, ransomware, data theft and loss of operational access.
No clear owner, policy, audit trail, approval threshold, incident route or human review standard.
A practical governance framework
The answer is not to reject innovation. Technology can reduce repetitive work, improve access to information and help firms organise complex material. The ethical task is to match the control to the risk. A low-risk administrative use does not require the same supervision as drafting a court document, analysing evidence or processing highly sensitive client data.
A defensible framework begins before procurement. The firm should identify the intended use, the information involved, the likely failure modes and the person accountable for the decision. Contractual assurances from a supplier are relevant, but they do not replace the firm’s own understanding of the system.
Controls should then follow the work. Staff need clear limits on what may be entered into a system. Material outputs need proportionate verification. High-risk decisions should retain meaningful human review. Security measures should include access management, updates, multi-factor authentication, backups and an incident plan. The firm should also be able to explain its decisions and demonstrate that supervision was real rather than nominal.
Specify the task, the information involved, the decision affected and the level of client or court risk.
Assess accuracy, security, retention, access, supplier terms, known limitations and failure modes.
Set permissions, verification standards, escalation thresholds, supervision and incident procedures.
Record the assessment, policy, training, reviews, material decisions and action taken when problems arise.
Source anchors
These official sources support the modern regulatory and information-governance framework. They do not apply to the fictional Victorian scenarios.
The fundamental duties concerning rule of law, public trust, independence, honesty, integrity and client interests.
The current duties concerning competence, supervision, accountability, confidentiality and safeguarding money and assets.
Guidance on protecting and not misusing current and former client information, including after a retainer ends.
Regulatory context for cyber risk, client assets and the need for firms to understand and mitigate their specific threats.
Data-protection guidance and risk-assessment resources for organisations using AI systems.
The closing point
The machines in the fog are fictional. The governance problem is not. Each generation produces tools that appear to shift responsibility away from the individual professional and into a system. Legal ethics resists that drift. A solicitor may use an advanced tool, but must still understand the work, protect the client, verify what matters and remain able to justify the decision.
Legal technology decision point
Get a free written assessment of the evidence route
Legal Lens can structure a preliminary written review where technology, legal service delivery and client harm intersect: the workflow, source documents, professional duties, evidence gaps and available route.
Identify the tool, the information used, the output relied upon, the human checks and the practical consequence.
Separate service, conduct, data protection, cybersecurity, negligence, insurer and court issues.
Technology use, professional duty, information risk, harm and route options.
The records needed before complaint, notification, escalation or claim analysis.
Independent Legal Lens consultancy. Legal Lens is not a regulated solicitors’ firm. A preliminary assessment is not a substitute for regulated legal advice where that is needed.