Claudia Santos Cruz^[1]
Morais Leitão Legal Circle/MDR Advogados, Lisbon

Ethics beyond automation

Technological competence is increasingly viewed as part of professional responsibility, although jurisdictions differ in how explicitly they codify it. In England and Wales, the Bar Standards Board (BRB) issued AI guidance in May 2026^[2] requiring barristers to maintain competence, supervise work appropriately and ensure that all outsourced or technology-assisted work complies with core duties such as independence, confidentiality and accountability. This is consistent with the broader professional principle that responsibility cannot be delegated.

The EU AI Act,^[3] the world’s first comprehensive AI regulation, classifies AI used in the ‘administration of justice’ as high-risk, imposing obligations on both providers and users – including law firms. Core obligations for high-risk systems are due to come into force in December 2027, with penalties for non-compliance of up to €35m or seven per cent of worldwide annual turnover. The Act’s extraterritorial reach means that law firms outside the EU whose AI outputs are used in the EU to assist the administration of justice^[4] are potentially within its scope.

However, technological systems and competence should not be confused with core professional legal competencies. Legal practice remains grounded in judgment, ethical reasoning and accountability – capacities that cannot be delegated to systems, regardless of their sophistication.^[5]

Traditional legal skills should be preserved and defended. Critical reasoning, ethical judgment and professional independence become more important, not less, in an AI-driven profession.

The more technology automates routine work, the more lawyers should be valued for the capabilities that AI cannot replicate.

Accountability in AI practice

The ethical boundary in the use of AI is not defined by whether AI is used to generate legal work, but by whether the lawyer continues to exercise meaningful critical assessment of the output. In other words, whether lawyers can use AI without compromising judgment and accountability.^[6]

Legal professionals have always relied on auxiliary tools: textbooks, case law, databases, expert witnesses, trainees and associates. These do not dilute professional responsibility because the lawyer remains intellectually accountable for the outcome. AI should be understood through the same lens as traditional delegation within a law firm: just as a partner cannot rely on a junior associate’s draft without review, a lawyer cannot rely on AI-generated material without independent scrutiny.

In this sense, AI does not introduce a fundamentally new ethical problem but rather amplifies an existing professional risk: the tendency under time pressure to substitute convenience for diligence. Responsible use of AI involves delegation of tasks for efficiency while retaining full responsibility for verification, including checking authorities, testing assumptions and considering alternative interpretations. Ethical failure arises where the lawyer effectively outsources reasoning itself and treats outputs as settled answers rather than material requiring evaluation.

Clients expect lawyers to assess risk, exercise discretion and assume responsibility for outcomes, functions that cannot be transferred to automated systems. Accordingly, liability for AI-generated errors, such as fabricated legal authorities in pleadings, rests with the lawyer rather than the developer, system or user interface, in the same way it would for errors introduced by trainees or paralegals.

Ultimately, AI does not alter the structure of professional accountability. While it increases efficiency, it also heightens the risk of intellectual complacency, making it easier to bypass rigorous analysis. The core ethical concern is therefore not technological risk but the exposure of professional character.

Less is more

The rise of professional social media engagement raises unresolved ethical tensions regarding personal branding, authority and professional conduct.^[7] Unlike regulated legal submissions, digital platforms operate without procedural safeguards, editorial controls or contextual constraints. Statements may be taken out of context, redistributed and reinterpreted indefinitely, creating reputational risk particularly where public commentary may later be scrutinised in adversarial or regulatory contexts.

Overexposure may encourage lawyers to prioritise commentary and immediacy over careful analysis and ethical reflection. In this environment, younger lawyers risk becoming commentators before they become trusted advisers, and personalities before they become professionals.

‘Performative lawyering’,^[8] driven by visibility metrics, further shifts incentives from correctness and careful judgment to speed and attention. Authority is no longer earned solely through experience, achievement or peer recognition; it is increasingly equated with visibility, engagement and algorithmic amplification. The most careful lawyer in any jurisdiction may have no online presence at all, while the most visible lawyer may simply be the most visible. Credibility is not the same as visibility.

These dynamics also contribute to a gradual blurring of personal and professional boundaries. Lawyers may regard social media posts as informal personal commentary, yet others may interpret them as legal advice, firm positions or indicators of bias. This can affect client relationships, raise concerns about independence and conflicts of interest and risks past statements subsequently being used to challenge credibility or impartiality. The core issue is not only what is said, but the inability to control how it is interpreted once it enters the public domain.

Professional regulators, including the Bar Standards Board (BSB), have begun to acknowledge these risks, emphasising that online behaviour falls within the scope of professional conduct rules, including duties relating to integrity and public trust.^[9]

Against this background, a limited professional presence mitigates risk, preserves clearer boundaries between personal and professional spheres and reduces exposure to misinterpretation, reputational volatility and ethical ambiguity. It ensures that reputation is built on more durable foundations: the quality of work, the trust of clients, the respect of peers and contributions made through institutions rather than personal promotion.

This is not an argument against social media in principle. Used responsibly, it can educate, inform, facilitate professional dialogue and improve access to legal knowledge. However, the profession should be cautious about treating visibility as an implicit obligation or a marker of professional worth.

Learning the law

Many firms are currently reassessing recruitment models and reducing trainee or junior lawyer intake on the assumption that AI can perform a significant portion of their work.

For generations, lawyers acquired expertise through repetition. They reviewed documents line by line. They analysed contracts clause by clause. They learned to identify inconsistencies, anomalies and risks through direct engagement with the material.

AI has created unprecedented efficiency in document review, due diligence and contract analysis, but the long-term risk is deskilling, particularly for junior lawyers. When an experienced lawyer reviews a contract line by line, subtle anomalies may be detected. This engagement trains the next generation to identify inconsistencies, evaluate risk and exercise judgment and accountability.

Junior lawyers are not merely document producers, they are the foundational tier of legal development. They become the future senior associates, partners, judges, regulators and leaders of the profession. They are also often the first layer of quality control and factual verification within a matter. If firms reduce training and verification opportunities in favour of efficiency, they risk reducing the future talent pipeline and weakening both quality control and the profession’s future capacity for independent legal reasoning.^[10]

A profession that systematically reduces opportunities for foundational training today may discover, ten years from now, that it has fewer lawyers capable of independently validating the outputs generated by increasingly powerful systems.

Current tendencies may end up creating professionals who can operate technology but lack the experience necessary to challenge it. The consequences could be significant with more errors and reduced institutional knowledge.

The cost of shortcuts

Clients are not external observers of AI adoption; they are active participants in shaping its trajectory. The integrity of the legal profession will therefore depend not only on lawyers defending professional limitations, but also on clients recognising that certain safeguards exist not for the lawyer’s benefit, but for their own protection.

Intensified economic pressure has resulted in clients demanding faster delivery and lower fees. While commercially understandable, they can unintentionally distort professional conduct by incentivising reduced verification and increased reliance on automated outputs.

Within firms, these pressures manifest in predictable ways: junior lawyers may become increasingly dependent on AI-generated drafts, senior lawyers may shift from interrogating outputs to merely reviewing them, and firms may reduce verification standards in order to remain competitive. In this context, verification becomes one of the first safeguards to erode, because it is largely invisible to the client.

From a legal risk perspective, verification is not inefficiency but a control mechanism. Legal services are not merely production exercises; they are risk management systems. When speed is prioritised over technical scrutiny, client exposure increases.

This dynamic is particularly evident in due diligence exercises. Where hundreds of contracts would traditionally be reviewed manually, AI tools can now extract key provisions in minutes. However, if a critical clause is misidentified, the consequences may only become apparent after completion of the transaction. What appears to be efficiency at the front end can therefore translate into failed transactions, unexpected liabilities or years of litigation. And the irony is that the client who demanded lower fees and faster delivery may ultimately bear that cost which is likely to far exceed the initial savings.

Treating legal services as a commodity inevitably increases risk, because legal work cannot be reduced to production output without affecting the quality of judgment and supervision. The assumption that legal work can always become faster, cheaper and equally reliable may therefore prove to be a costly misconception of the AI era.

Clients should understand that legal services operate within a framework of risk allocation, not simply cost efficiency. The integrity of that framework depends on maintaining a balance between speed, accuracy, supervision and legal judgment. The integrity of the profession and the protection of clients alike depend on ensuring that economic pressures do not quietly erode professional standards while technology disguises the shift.

Conclusion

The ethical divide in an AI-enabled profession will not be between users and non-users of technology, but between lawyers who use tools to enhance judgment, accountability and risk management, and those who allow technology to substitute these. Similarly, visibility and personal branding cannot supplant competence, discretion and sustained professional reputation. Technology can reward efficiency, but it cannot bear responsibility. Lawyers must use AI intelligently, supervise its outputs, train future generations and preserve the professional standards that underpin trust in the legal system.