Digital Transformation, Labor Data and Strategic Litigation in Brazil: A Practical Matrix for Legal Innovation
Digital transformation has reached labor and employment law through a less spectacular, but far more consequential route than most public debates suggest. It is not only about replacing human work with machines or predicting the end of traditional legal practice. Its real impact lies in the way companies hire, allocate, monitor, remunerate, negotiate with and eventually dismiss workers. In Brazil, where labor relations are shaped by statutory rules, collective bargaining, intense litigation and a highly specialized Labor Court system, this transformation requires a particularly careful reading. Brazilian labor law is not a field in which technology can be assessed only by speed or visual sophistication. A platform may summarize lawsuits, classify documents, generate dashboards or automate procedural steps. These functions may be useful. But usefulness is not the same as legal transformation. The more relevant question is whether technology improves the quality of judgment, the traceability of evidence, the consistency of strategy and the ability to act before risk becomes litigation. Otherwise, one may simply be looking at a well-organized workflow with a modern interface, something valuable, but not necessarily revolutionary. The Brazilian context makes this distinction essential. Labor risk is rarely located in one document or one lawsuit. It usually emerges from the interaction between employment contracts, payroll practices, job descriptions, working-time records, collective bargaining agreements, internal policies, benefits, occupational health and safety documents, HR systems and the actual way work is performed. A serious labor analysis must therefore move across law, operations, data and evidence. This is especially true in corporate transactions. Mergers, acquisitions, spin-offs, incorporations and intragroup reorganizations have always required labor due diligence. The traditional review remains necessary: employment contracts, headcount, pending lawsuits, severance liabilities, collective bargaining agreements, succession rules and payroll exposure. In Brazil, the Consolidation of Labor Laws protects employees against changes in ownership or corporate structure that could affect their rights. But in modern transactions, this is only the beginning of the analysis. The first step is to understand the workforce architecture. Each employee must be mapped according to legal entity, workplace, actual duties, reporting line, cost center, business unit, corporate purpose of the employing company and applicable union framework. In practice, the formal employer on payroll may not fully reflect the operational reality. This matters because allocation affects union classification, collective agreements, salary floors, benefits, working-time rules, profit-sharing arrangements and potential claims involving group companies or corporate succession. The second step is the collective bargaining map. Brazil’s labor system gives central importance to collective bargaining agreements and company-level agreements. They may regulate salary floors, meal and food allowances, health plans, overtime, bank of hours, shifts, profit-sharing, union contributions, stability rules, occupational conditions and penalties for non-compliance. In a corporate transaction, transferring employees from one company to another may alter the applicable union, territorial basis, bargaining date or collective instrument. A merely formal comparison of documents is insufficient. The legal team must understand which clauses affect cost, which affect operations, which require system parameterization and which may demand negotiation before integration. The third step is the compensation and benefits matrix. Brazilian labor litigation often arises from differences that were not properly documented at the time they were created. Distinct salary bands, bonus formulas, commissions, benefits, allowances, vehicles, remote-work policies and eligibility criteria may become sensitive when employee populations are integrated. The relevant question is not simply whether differences exist. The question is whether they are legally justified, objectively documented and operationally sustainable. Equal pay risks, discrimination claims, payroll misclassification and disputes over incorporated benefits are often born from the absence of a clear explanation. The fourth step is the review of contracts and internal policies. Employment contracts, amendments, confidentiality clauses, intellectual property provisions, mobility rules, remote-work arrangements, compensation policies, expense policies, performance evaluation criteria and disciplinary procedures should be compared with actual practice. A policy that looks adequate in isolation may become fragile when read against collective bargaining obligations or payroll records. Conversely, a contractual clause may be formally valid but irrelevant if the company’s operational practice contradicts it. The fifth step is litigation analytics, or jurimetrics. In Brazil, where labor litigation is both voluminous and technically specialized, data can support better decisions in individual and collective disputes. A well-structured database can classify claims by subject, claimant profile, court, region, procedural stage, amount claimed, amount provisioned, evidence available, settlement history, judgment pattern and recurrence of factual causes. This allows companies to distinguish isolated litigation from systemic risk. For individual cases, jurimetrics can improve settlement strategy, provisioning, witness preparation and procedural prioritization. It may show, for example, that certain claims have higher exposure when specific documents are missing, when working-time records are inconsistent, or when a particular factual pattern repeats across units. It can also help identify cases in which early settlement is more rational than defensive litigation, not because of generic risk aversion, but because the data indicates a poor combination of evidence, jurisdiction, claim type and expected cost. For collective matters, the use of data is even more strategic. Union negotiations, public civil actions, mass claims and recurring disputes require a broader view than case-by-case defense. Data can reveal whether a certain claim is concentrated in one region, one job family, one manager group, one payroll item, one benefit policy or one timekeeping practice. It can also support collective bargaining by showing the financial and operational impact of different scenarios, including harmonization of benefits, changes to working-time rules, profit-sharing models and transition clauses. This is where technology may create real value. It can organize large volumes of information, extract clauses from collective instruments, compare payroll data, identify inconsistencies, group lawsuits by theme, detect recurring factual causes and generate risk dashboards that connect litigation to business decisions. But the technology must remain subordinate to legal method. A dashboard is not a legal opinion. A prediction is not a strategy. A cluster of similar lawsuits is not, by itself, a diagnosis. The lawyer must still ask the hard questions: what is the evidence, what is the applicable collective instrument, what is the procedural
The New Era of Labor Compliance in Brazil: How the PGR and NR-01 Transformed Corporate Mental Health Management
Brazil has a comprehensive regulatory framework for protecting the health and safety of its workforce. These rules are created and updated by the Ministry of Labor and Employment to prevent occupational accidents and illnesses. Known as Regulatory Standards (Normas Regulamentadoras – NRs), compliance is mandatory for any company employing workers under formal employment contracts (carteira assinada). To use a game analogy, the NRs serve as the game’s rulebook, where the core objective is to protect workers’ health and safety. There are currently 38 active NRs covering a wide range of topics, such as safety standards for working at heights (NR-35), health services guidelines (NR-32), sanitary and comfort conditions in workplaces (NR-24), and protocols for handling explosives (NR-19), among others. Recently, a highly significant shift occurred within this regulatory architecture. NR-01, which establishes the foundation for the entire occupational health and safety system in Brazil, outlines the general provisions, scope of application, definitions common to all NRs, and the guidelines and requirements for occupational risk management and preventive measures. The Scope of Occupational Risk Management (GRO) Recognizing the growing importance and complexity of mental well-being in workers’ daily lives, the updated NR-01 now explicitly integrates psychosocial risk factors into the scope of Occupational Risk Management (Gerenciamento de Riscos Ocupacionais – GRO). The GRO establishes a systematic framework for hazard identification, risk assessment, and control. It must be seamlessly integrated with other medical initiatives (such as the Occupational Health Medical Control Program – NR-07), accident analysis, and emergency preparedness. The primary objective of the GRO is the prevention and mitigation of workplace risks, and NR-01 mandates its implementation across all of an organization’s business establishments. The core workflows of the GRO must be formalized into an Occupational Risk Management Program (Programa de Gerenciamento de Riscos – PGR), which is defined as: “A coordinated set of actions by the organization designed to achieve occupational risk prevention and management objectives, formally documented.” — Subitem 1.5.3.1.1 of NR-01 Under these regulations, implementing a PGR is mandatory for each business location and must cover all activities performed there. At a minimum, it must include an occupational risk inventory, an action plan, and the criteria used for risk assessment within the GRO/PGR framework. To ensure full compliance with NR-01, organizations must also document additional supporting records, including: Workplace accident and illness analysis reports (subitem 1.5.5.5.2); Implementation logs for preventive measures (subitem 1.5.5.3.1); Planned monitoring records for the performance of preventive measures (subitem 1.5.5.3.2); Evidence of simulated emergency response drills (subitem 1.5.6.3.1); Training and capacity-building records as prescribed by the NRs. The operational flow of this documentation can be simplified into a continuous cycle: Identify risks – Assess – Control – Monitor. The Transition from PPRA to PGR: Integrating Ergonomics and Psychosocial Factors The most substantial update lies in the scope of the risk assessment. The former PPRA (the program replaced by the PGR) was strictly limited to environmental hazards (physical, chemical, and biological risks). However, under the updated NR-01, the PGR must also encompass risks stemming from ergonomic factors, which explicitly include work-related psychosocial risk factors. This regulatory shift is milestone-heavy because it legally recognizes the central role of mental health in overall worker well-being. This alignment is further reinforced by NR-17, which embeds psychosocial risk factors directly into ergonomic management—the central focus of that standard. To achieve NR-17’s objective of adapting working conditions to the psychophysiological characteristics of workers, the standard mandates two complementary methodologies: Preliminary Ergonomic Evaluation (Avaliação Ergonômica Preliminar – AEP): which is mandatory; Ergonomic Workplace Analysis (Análise Ergonômica do Trabalho – AET): required only under specific operational circumstances. Through the AEP, risk assessments for ergonomic and psychosocial hazards can be conducted using qualitative and participatory approaches, integrating them directly into the PGR’s baseline hazard identification process. This implementation must always be multidisciplinary and multi-professional. Crucially, workers must have a voice in this process. To facilitate open dialogue, companies must cultivate an environment of psychological safety and conduct surveys anonymously, ensuring employees do not feel intimidated by fears of exposure or reprisal. The Organizational Roots of Psychosocial Risks It is vital to emphasize that the root source of psychosocial risks does not reside within the individual worker, but rather within the organization of work itself. This includes deficiencies in task design, personnel management, and organizational workflows. Left unmanaged, these systemic issues trigger severe psychological, physical, and social health consequences, including workplace stress, burnout, Work-Related Musculoskeletal Disorders (WMSDs/DORT), and depression. Consequently, identifying psychosocial risks requires evaluating work organization—meaning how tasks and activities are structured, distributed, and coordinated within the environment. When evaluating these factors, organizations must analyze which elements of the work activity act as stressors with the potential to cause injury or health deterioration. The focus is not on diagnosing individual symptoms or capturing subjective “moods,” but on auditing objective working conditions, identifying operational stressors, and evaluating environmental and systemic flaws. Once psychosocial risks are identified, companies must assess and classify them. The risk level is calculated by combining the severity of potential injuries or health impairments with the probability of occurrence, as outlined in subitem 1.5.4.4.2 of NR-01. With the diagnosis complete, the necessary preventive measures must be integrated directly into the PGR Action Plan. The Current Landscape: Compliance, Sustainability, and the Future of Governance The urgency of this landscape—which was already alarming in 2021 when World Health Organization (WHO) data indicated that 359 million people globally suffered from anxiety disorders—has consolidated over the past five years as one of the defining corporate governance challenges of our time. In 2026, with the maturation of hybrid working models and accelerated technological transformation, it has become undeniable that mental well-being can no longer be treated as a purely individual responsibility or an auxiliary corporate benefit. It is, fundamentally, an outcome of organizational design. Therefore, the inclusion of psychosocial risks within the PGR and NR-17 represents far more than a bureaucratic milestone; it is a definitive turning point for corporate compliance and ESG sustainability. By shifting the regulatory focus away from employee symptoms and toward
The Taxation Omnibus and The Return of Competitiveness: What The EU’s Simplification Turn Asks of Anti-Abuse
When the European Commission adopted its Tax Simplification Package on 24 June 2026, the presentation was almost entirely about relief: some eight billion euros a year in lower compliance costs, a lighter withholding regime, fewer overlapping calculations, a research allowance to draw investment to the continent. The headline was decluttering, and on that measure the package delivers. Yet read against the decade that produced it, the Omnibus is less a housekeeping exercise than a change of direction. For most of the post-BEPS period the centre of gravity in EU direct tax was anti-abuse, as successive directives layered controlled-foreign-company rules, interest limitations, hybrid-mismatch provisions and a general anti-abuse rule onto national systems, each meant to close a gap. The Omnibus reverses the reflex: its animating question is no longer how to prevent leakage but how to keep capital, financing and research inside the Single Market, and that reversal is where the interesting problems sit. What The Package Actually Does The Simplification Package comprises two instruments: a draft Omnibus Directive amending six tax directives, and a recast of the Directive on Administrative Cooperation consolidating the DAC framework into a single text. Four measures carry most of the weight for corporate groups. The most consequential is the treatment of withholding taxes. The Omnibus would exempt cross-border payments of dividends, interest and royalties between EU companies from source taxation, attacking the conditions, not merely the rate. The minimum-holding thresholds that currently gate the Interest and Royalties Directive and the Parent-Subsidiary Directive would go, as would the prior-authorisation procedures used to verify entitlement before payment; a self-assessment model would replace them, with the FASTER refund mechanism as a backstop where eligibility cannot be confirmed upfront. The Parent-Subsidiary Directive would also extend to pension institutions. On the Commission’s own figures these account for the bulk of the savings, which tells you where the package’s centre really lies. The second measure reshapes the interest limitation rule. The thirty-percent-of-EBITDA cap becomes uniform and mandatory, the three-million-euro de minimis safe harbour becomes a floor rather than a ceiling, indexed to inflation so that it can only rise, and genuine third-party and market financing is carved out where the borrowing funds the taxpayer’s own activities rather than on-lending within the group. Optional elements that fragmented implementation, the group-escape and carry-forward mechanisms, become mandatory. The defence sector is temporarily excluded, as much a political signal as a tax measure. A uniform mandatory cap also overrides deliberate national choices: the Netherlands, among others, had set its threshold below thirty percent, and would have to loosen a rule it chose to keep tight. The third measure is the one practitioners should watch most closely, because it is where simplification and sovereignty collide. The Omnibus removes the overlap between the ATAD controlled-foreign-company rules and the Pillar Two global minimum tax by exempting groups within scope of Pillar Two from the CFC regime altogether. The logic is clean: a group already computing top-up tax on low-taxed subsidiaries should not also run a parallel CFC calculation on the same income, a duplication the Commission puts at around a hundred and sixty million euros a year. The two CFC models then collapse into one, the passive-income approach made mandatory and divergent national variants barred. The fourth measure points in a different direction entirely. A new research-and-development allowance, grafted onto the ATAD, would give full and immediate expensing of qualifying tangible R&D assets as a binding minimum standard across the Union. That an anti-avoidance directive should now house an investment incentive captures the whole shift: the instrument built to protect the base is being asked to grow it. Running in parallel, the DAC recast narrows DAC6, carving out Pillar Two groups from cross-border-arrangement reporting and deleting hallmarks judged to generate more noise than signal. The Problem Underneath The Relief None of this is straightforwardly deregulatory, nor a retreat from anti-abuse. The general anti-abuse rule is not weakened; it is extended, reaching beyond corporate tax to withholding taxes and to the Pillar Two top-up taxes themselves, closing an uncertainty lingering since the minimum tax arrived. A subject-to-tax safeguard guards against the withholding exemptions producing double non-taxation where the recipient is taxed nominally at zero. The architecture of protection remains; what changes is its calibration. The Commission’s wager is that several anti-abuse rules were addressing risks that Pillar Two now covers, so that maintaining both is not prudence but duplication. Yet the same package that widens the GAAR narrows anti-abuse elsewhere. The imported mismatch provisions of ATAD, which denied a deduction where a payment financed a hybrid mismatch further down the chain, are removed outright, the Commission judging them too complex and poorly targeted. Unlike the CFC change, no other instrument steps into the gap: the rule goes not because something else does its work, but because it was hard to apply. That asymmetry, anti-abuse expanding on one front and contracting on another within a single directive, is the clearest sign that usability, not protection, is now the organising principle. That wager is defensible, but it is a wager, and it exposes the deeper tension in the package. Simplification in EU direct tax is not a technical act: every rule the Omnibus makes uniform is a rule a Member State can no longer tune to its own base, and every carve-out is revenue foregone somewhere. The CFC exemption is the clearest case. Removing the overlap with Pillar Two sounds unanswerable until one remembers that Pillar Two, by design, raises little revenue in many jurisdictions, while CFC charges can raise real amounts. States that built their regimes on the transactional Model B, Ireland, Malta and the Netherlands among them, are asked to abandon it for a mandatory passive-income model and to surrender a working revenue tool on the theory that a lower-yielding one has superseded it. It is no surprise the CFC changes are expected to be among the most contested at Council. This is the structural difficulty the Omnibus cannot draft its way around. Direct tax measures require unanimity
The Value of Tax Lawyering in an Age of Quick Answers
Information moves fast. Judgment doesn’t. The starting point for a tax lawyer’s work has changed. For a long time, much of this work began with the search for information — finding the relevant law, mapping out case law and legal doctrine, and comparing interpretations. Today, clients often come to the meeting having already researched the topic using an artificial intelligence tool. What they expect from their lawyer has changed. And it is this change that is worth reflecting on. The most obvious reading of this scenario is that tax lawyering will become simpler and, perhaps, unnecessary. If information is more accessible, why would it still be necessary to rely on experienced professionals? The answer is straightforward: in tax law, information is not the same as an answer. And an answer is not the same as professional judgment. Technology has brought clear benefits. It saves hours previously spent on background research, organizing information, and comparing and summarizing texts and regulations. But there is a point where it stops. And it is precisely at that point that the most difficult work begins: how to advise a client who does not want just a quick diagnosis of the problem, but guidance on what to do — in an environment marked by constant legislative changes that are not always clear, by divergent interpretations of the same issue (sometimes adopted by the same court or authority), and by shifts driven by the political, economic, and social context, both locally and internationally. Tax issues are rarely just matters of research. The client doesn’t always want to know if there’s a legal argument for a particular course of action. They want to understand whether that course of action makes sense at that moment, within that structure, in that sector, and in that commercial context. They want to know if it’s worth moving forward, or if what looks like a technical alternative today could turn into a tax assessment tomorrow, years of litigation, or a headline no one wants to read. This kind of guidance isn’t readily available anywhere. It is built through analysis of the specific case — looking at what has already been done, what can still be done, and what may not be evident at first glance. In some tax systems, there is a significant complicating factor: the outcome in similar situations can vary depending on the strategy adopted by the taxpayer. In other words, similar cases may have different outcomes depending on the path chosen. That path may include paying the tax to avoid assessments, even where the amount due is uncertain; waiting for the statute of limitations to expire as a form of tacit validation by the tax authorities; contesting the matter through administrative or judicial channels; pursuing the dispute at trial or on appeal; adopting a more defensive or more assertive posture; participating in installment programs or settlements with the tax authorities; or pursuing other ADR mechanisms, among others. This shows just how untenable it is to think that tax lawyering can be fully standardized. What is at stake, often, is not merely the legal argument itself, but the way it will be used, defended, and sustained over time. And it is precisely in this realm — where the rule exists, but the correct answer depends on the context — that the value of professional judgment becomes most evident. Over time, tax lawyers learn to recognize patterns, to distinguish between a technically clever argument and a position that is truly tenable, and, above all, to understand that not everything that is possible on paper is advisable in practice. This is an insight built slowly, from years of watching how these issues evolve, and it is precisely this kind of understanding that is difficult to extract from a database. When there is so much talk about artificial intelligence in the legal world, the most interesting question may not be whether AI can help. It certainly can. The real question is different: what remains profoundly human in tax lawyering? The answer seems to lie in the ability to exercise discernment in the face of complexity. The best tax lawyers are not merely those who identify a legal argument. They are the ones who can weigh risks, foresee consequences, perceive nuances, anticipate how authorities and courts are likely to react, and transform all of this into clear and responsible guidance. Ultimately, clients aren’t just looking for information. They’re looking for the confidence to choose a path. They’re looking for someone to help them understand where the risks lie, which paths are more solid, and how to move forward with greater clarity. That confidence doesn’t come from the speed of the response. It comes from the quality of the professional judgment behind it. When everyone has access to information, and preliminary answers are generated in seconds, what matters most is who knows how to ask the right questions and identify what is truly at stake. And when the landscape is one of uncertainty — with significant financial, reputational, and operational impacts — it remains essential to rely on professionals capable of exercising sound judgment. Perhaps tax lawyering has never really been about finding information. It has always been about what to do with it. Technology has changed the starting point; it has not changed the destination. And it is at the destination that professional judgment continues to make the difference. About the Author: Gabriela Mariano Schunck, Tax Counsel, Knowledge Management, Latin America at Trench Rossi Watanabe (in strategic cooperation with Baker McKenzie) Gabriela Mariano Schunck is a tax lawyer with over 20 years of experience advising Brazilian and multinational companies on complex tax matters. She is Counsel in the Tax Practice at Trench Rossi Watanabe, leading the Tax Knowledge Management area in Brazil and across key Latin American jurisdictions under the firm’s strategic cooperation with Baker McKenzie. Her expertise includes direct and indirect taxation, tax compliance, planning, M&A transactions, and corporate reorganizations. Gabriela has also worked at Big Four firms and as a partner at
Compliance is the Name of the Game
The financial services sector has suffered major disruptive changes in the past years, with the increasingly complex regulatory and compliance landscape, the accelerated advancement of technology, the constant evolution of cyber threats and each time more sophisticated money laundering schemes intended to deceive banking institutions. Billions of dollars from drug trafficking and illegal activities attempt to be laundered and recycled back to the economy through financial institutions, despite rigorous AML/CFT bank controls in place. According to the United Nations Office on Drugs and Crime, the estimated amount of money laundered globally in one year represents 2 – 5% of global GDP, or $800 billion – $2 trillion in current US dollars. However, these figures may be underestimated due to the difficulties of detecting misuses of the financial system. In this regard, US has issued several bills, directives and orders aimed at deterring drug-related financial crimes. Amongst them: i) The Fentanyl Sanctions Act (S.1044), enacted on December 20, 2019, decreeing the imposition of sanctions with respect to foreign traffickers of illicit opioids, ii) the Fentanyl Eradication and Narcotics Deterrence Off Fentanyl Act ( S.1271) enacted during President’s Biden administration, as part of a national security set of laws, iii) the Halt All Lethal Trafficking of Fentanyl Act (H.R. 27), passed during the current administration of President Trump, classifies fentanyl-related substances as Schedule I under the Control Substances Act, and iv) the Executive Order 14157 (E.O. 14157) enacted by President Trump, redefining international drug cartels as Foreign Terrorist Organizations (FTOs). Within this context, on June 25, 2025, the U.S. Treasury’s Financial Crimes Enforcement Network (FinCEN) took an unprecedent action preventing U.S. covered financial institutions from engaging in fund transfers related to 3 mid-size Mexican financial institutions, cutting them off from the US financial system and thus, globally. These institutions posed serious money-laundering concerns. The effects were devastating for the 3 sanctioned entities; they were forced to enter liquidation or acquisition processes. Mexican authorities were able to contain a systemic risk while maintaining market stability. The Department of the Treasury has the authority to take actions related to financial institutions, transactions, or types of accounts that (1) involve a non-U.S. jurisdiction, and (2) are of primary money laundering concern in connection with illicit opioid trafficking or terrorism. Mexico’s banking industry has now learned that no compliance controls are ever enough. Preserving global financial integrity requires the undertaking of collective efforts by government agencies and other stakeholders, sharing experiences and best practices, join collaboration with industry peers, homologation of rules and standards and combat via evolving legislation and other preventive measures. After the FinCEN order, most Mexican banks have reinforced their AML/CFT controls in alignment with US and Europe regulations. Bankaool’s incursion in the FX market surged after the FinCEN order, so the entity has always been aligned to such regulations and keeps close collaboration with both regulators and industry peers. About the Author: Elizabeth Ceballos, Head of Financial Institutions at Bankaool Elizabeth Ceballos is a seasoned lawyer admitted to practice in 1987, with extensive experience in banking, energy, and public law. She currently serves as Head of Correspondent Banking at Bankaool, where she manages domestic and cross-border banking relationships, with deep expertise in KYC processes and international regulatory compliance. Previously, as Oil & Gas Global Leader at EY, she advised clients on complex regulatory and contractual matters related to Mexico’s Energy Reform and major energy projects. Her experience also includes public-private partnerships, investment projects, and contract negotiations, making her a trusted advisor in highly regulated and cross-border business environments.
COMPLIANCE WAS NEVER THE ENEMY. CONFUSION WAS.
The Conversation That Started Everything There is a conversation I have had more times than I can count. A business owner, a founder, an operations lead, someone running a real company with real employees and real clients, sits across from me and says some version of the same thing: “I know we probably should be doing more on compliance. We just don’t really know where to start.” That sentence carries more than uncertainty. It carries the weight of a system that was never designed to explain itself to the people who need it most. After years working inside Big Tech, where legal and compliance teams have the budgets, the headcount, and the institutional muscle to stay on top of regulatory change, I kept noticing something that troubled me. The companies getting into trouble were not the ones trying to evade the rules. They were the ones who genuinely did not understand them. Not because they were careless. Because no one had ever translated the rules into language they could act on. That observation became a thesis: compliance failures stem from confusion, not dishonesty. And that thesis changed everything about how I think this problem should be solved. The Gap Between Documentation and Protection The traditional approach to compliance is built on documentation. Write the policy. Train the staff. File the record. Repeat annually. This approach was designed to produce what I would call regulatory defensibility, the ability to demonstrate, when an auditor or regulator comes knocking, that the right paperwork exists. And it works, in the sense that it produces paperwork. What it does not always produce is genuine protection. The gap between having documentation and understanding what that documentation requires of your people, in real situations, in real time, is where most compliance failures live. A policy that no one has read, written in legal language for a legal audience, sitting in a shared drive no one visits, is not a compliance program. It is a file. This distinction matters enormously for small and medium-sized businesses, which make up most of the productive economy across Latin America and everywhere else. These are companies that cannot staff a legal department. They cannot retain outside counsel for every regulatory question. They are running their businesses, and compliance sits somewhere on the to-do list between urgent and important, perpetually deferred until it becomes a crisis. Why Templates Are Not the Answer The standard response to this problem has been to offer simplified templates, generic checklists, or scaled-down versions of enterprise compliance programs designed for organizations ten times their size. None of these solve the problem, because the problem is not a lack of documents. The problem is a lack of clarity. What a business owner needs is to understand what applies to them, specifically, and why. Not a survey of everything the law could theoretically require, but a clear answer to the question: given what my company does, where we operate, and how we are structured, what do I need to have in place? And then, once that question is answered, they need the actual documents. Not templates. Documents written for their business, their contracts, their risk profile, their industry. Where Technology Changes the Equation This is where technology changes the equation. Not because it replaces judgment, but because it distributes it. The analytical work that used to require a senior compliance attorney, the kind of thinking that synthesizes regulatory requirements across jurisdictions, identifies gaps in existing documentation, and builds a program around the specific needs of a specific organization, can now be done at a fraction of the cost and in a fraction of the time. Human judgment is still there. It is embedded in the design of the tools, in the questions the system asks, in the logic that connects a company’s answers to the outputs it receives. What changes is who can access it. This is not the same as saying AI replaces lawyers. It does not, and it should not. Compliance at its core is judgment discipline. AI can review thousands of documents, flag anomalies, and surface risks faster than any human team. What it cannot do is read the room. It cannot weigh the context that turns a policy breach into a genuine mistake versus deliberate misconduct. It cannot understand why a particular employee made a particular decision under a particular kind of pressure. The most effective compliance programs are the ones that use technology to ask better questions and free up human capacity to focus on the conversations that matter. Compliance as a Growth Foundation What I have seen, both in Big Tech and in the conversations, I have had with SMB founders since, is that the companies that get compliance right are the ones that treat it as infrastructure, not as insurance. They do not build compliance programs to survive an audit. They build them because a clear, well-documented set of internal controls is what allows you to move faster, enter new markets with confidence, close deals without stalling in legal review, and build the kind of trust with clients and partners that is genuinely hard to replicate. A business that can answer the question “how do you handle my data?” without hesitation is not just legally protected. It is commercially advantaged. A company that has its supplier agreements in order does not discover mid-contract that it has accepted liability it never intended to carry. A law firm that has built its quality management program to a defined standard does not scramble when an audit arrives. In each of these cases, the compliance infrastructure is doing something that goes far beyond regulatory compliance. It is enabling growth. The Regulatory Landscape Is Not Waiting The environment is also changing in ways that make this infrastructure more urgent. Ethics in AI is no longer a philosophical aspiration. It is becoming codified law. Across the European Union, across Latin America, and in jurisdictions where regulatory conversation is still developing, the obligations around how businesses use AI, how
The Accountability Gap: Why AI Liability Is Ultimately a Human Problem
Artificial Intelligence is rapidly reshaping the architecture of modern decision-making. Algorithms now influence who receives credit, who gets hired, which transactions are flagged as suspicious, what information reaches consumers, and increasingly, how legal, financial, and operational risks are assessed within organizations. In many sectors, AI has quietly evolved from a productivity tool into critical decision-making infrastructure, becoming embedded in processes that directly affect rights, opportunities, reputations, and economic outcomes. As this transformation accelerates, a deceptively simple question is emerging in boardrooms, regulatory agencies, and courtrooms around the world: who is liable when AI fails? While the question appears straightforward, it risks obscuring the deeper legal challenge. Artificial intelligence does not sign contracts, owe fiduciary duties, appear before courts, compensate injured parties, or assume legal obligations. Despite increasingly sophisticated and human-like outputs, AI remains a technological instrument rather than a legal actor. Consequently, the central issue is not whether artificial intelligence should be held accountable, but rather how legal systems should allocate responsibility among the individuals and organizations that design, train, deploy, supervise, commercialize, and benefit from these systems. The significance of this debate extends far beyond technology regulation. Artificial intelligence is forcing legal systems to confront fundamental questions regarding governance, accountability, and the allocation of risk in increasingly automated environments. More importantly, the real question is not whether AI becomes more powerful. Technological progress strongly suggests that it will. The real question is whether our legal institutions, governance structures, and regulatory frameworks will mature at the same pace. The Return of an Old Legal Question Every transformative technology eventually confronts the legal system with the same fundamental challenge: how should responsibility be allocated when innovation causes harm? Throughout history, the printing press, railroads, electricity, aviation, the internet, social media platforms, cryptocurrencies, and countless other disruptive technologies have generated legal uncertainty precisely because they altered existing patterns of economic activity and social interaction. Although the technologies themselves differed dramatically, the underlying legal problem remained remarkably consistent. Legal systems do not evolve in isolation. They adapt when technological developments alter the practical constraints under which societies operate or create economic incentives too significant for existing frameworks to ignore. The history of law is, in many respects, the history of institutional adaptation to changing realities. Yet law remains inherently conservative. Courts look to precedent, legislators build upon existing frameworks, and regulators often seek continuity rather than disruption. When confronted with genuinely novel technological challenges, legal institutions rarely invent entirely new doctrines from scratch. Instead, they attempt to understand emerging issues through analogies to established legal concepts. Artificial intelligence is likely to follow a similar trajectory. The challenge facing courts and regulators will not be the absence of legal tools, but rather determining which tools are most appropriate. Questions surrounding AI-related harm may be analyzed through the lenses of negligence, product liability, consumer protection, agency law, professional responsibility, contractual obligations, or data protection frameworks. As has occurred during previous technological transitions, existing doctrines will likely be adapted and refined rather than replaced altogether. While the actors participating in these disputes may differ from those involved in earlier technological revolutions, the principles used to allocate responsibility are likely to remain surprisingly familiar. AI Is Challenging the Assumptions of Private Law What makes artificial intelligence particularly significant is not merely its commercial impact or technical sophistication. Rather, it is the extent to which AI challenges assumptions that have quietly underpinned private law for centuries. Most legal systems were built upon a relatively simple premise: legally relevant decisions are made by human beings or by legal entities acting through human beings. Responsibility could therefore be traced to identifiable actors capable of exercising judgment, assuming obligations, and bearing legal consequences. Artificial intelligence increasingly complicates that assumption. In a growing number of narrow domains, advanced AI systems are already outperforming human experts. Sophisticated models can process enormous quantities of information, identify patterns invisible to human observers, and generate recommendations with levels of speed and accuracy that frequently exceed those of experienced professionals. This increased level of machine intelligence raises significant challenges for legal frameworks that have traditionally treated human judgment as the central element of legally relevant decision-making. The issue is not whether machines are becoming intelligent in a technical sense. The issue is whether legal systems remain capable of identifying responsibility when intelligence, decision-making, and operational control become distributed across increasingly complex technological ecosystems. In sectors such as healthcare, finance, insurance, employment, and legal services, algorithmic outputs increasingly influence decisions carrying substantial legal and economic consequences. As these systems become more capable and more deeply integrated into organizational processes, the boundary between human judgment and machine influence becomes increasingly difficult to define. The Emerging Accountability Gap One of the most significant risks associated with artificial intelligence is not technological failure itself, but the emergence of what may be described as an accountability gap. Traditional liability systems function most effectively when decision-making authority can be clearly identified and linked to a specific actor. Artificial intelligence complicates this process because responsibility becomes fragmented across multiple participants within a broader technological ecosystem. Model developers design and train the underlying systems, data providers contribute the information upon which those systems depend, technology vendors commercialize the resulting products, and organizations integrate them into business processes that ultimately affect employees, consumers, and other stakeholders. When harm occurs, each participant may plausibly argue that another actor exercised greater control over the final outcome. The result is a diffusion of responsibility that challenges the foundations upon which conventional liability frameworks have historically operated. This challenge becomes particularly visible when AI systems are used to approve loans, screen job applicants, detect fraud, assess insurance claims, or generate legal and financial recommendations. If the resulting outputs are discriminatory, inaccurate, misleading, or otherwise harmful, identifying responsibility becomes substantially more complicated than in traditional cases involving defective products or negligent professionals. The relevant question may involve flawed training data, architectural limitations, inadequate supervision, improper implementation, unreasonable reliance on automated outputs, or some combination of these factors. Artificial intelligence is
AI in Law is only as good as its knowledge foundations
Rethinking Knowledge Management as Strategic Infrastructure AI in Law is only as good as its knowledge foundations Rethinking Knowledge Management as Strategic Infrastructure Across law firms, in-house legal teams, and professional services organizations, AI adoption is accelerating rapidly. The promise is clear: faster work, improved efficiency, and enhanced competitiveness. Tasks are automated, documents are generated in seconds, and access to information appears instantaneous. Yet, outcomes remain inconsistent. Outputs vary, and quality still depends on human validation, requiring professionals to review, correct, and question machine-generated results. The issue is not technological capability, it is the absence of a reliable foundation. Artificial intelligence does not create knowledge, it processes it, recombines it, and scales it. Its effectiveness depends entirely on the quality and structure of the information it draws from. In law, where precision, consistency, and context are essential, this dependency is decisive. The implication is clear: AI is only as good as the knowledge behind it. Most legal organizations already hold vast amounts of knowledge: contracts, precedents, internal guidance, and accumulated expertise. However, possessing knowledge does not mean using it correctly. Much of it remains fragmented across systems, duplicated across teams, or embedded in individual experience. The problem is not lack of expertise, it is lack of structure. This inefficiency is often invisible. Information exists but is difficult to locate, validate, and leverage. When AI is introduced into such environments, it amplifies these gaps. Inconsistent inputs lead to inconsistent outputs, forcing professionals to spend time verifying rather than trusting results. This does not eliminate the need for human review. However, when the knowledge that feeds AI is accurate and well-structured, outputs become more consistent, require fewer revisions, and significantly reduce the risk of hallucinations. The difference is not the technology, it is the knowledge environment that supports it. This requires a shift in how Knowledge Management is understood. Traditionally positioned as a support function useful but secondary; in the era of AI, KM is not about storing documents, it is about transforming institutional knowledge into an operating asset. When this foundation is in place, the impact extends beyond efficiency and reshapes how work is performed. Repetitive tasks decrease, and drafting becomes faster and more consistent. Information is available when needed, allowing professionals to focus on higher-value work like strategy, client relationships, negotiation, and decision-making. This is where value is created: not in doing the same work faster, but in enabling a different use of expertise. Organizations that recognize this shift do not view AI as a cost-reduction tool. They see it as a capability enabler, one that scales expertise, standardizes quality, and creates operating efficiencies. Achieving this, however, requires deliberate action. Knowledge does not organize itself. It must be structured, governed, and continuously maintained. Organizations need clarity on what is aligned, what is current, and what can be reused. This requires common language, clear taxonomies, standardized templates, and disciplined content lifecycle management. The starting point is not technology, but questions: What knowledge do we have? Where does it reside? Is it consistent and current? Can it be trusted? Is it structured for AI to use effectively? These are not technical questions, they are strategic ones. They lead to initiatives such as knowledge audits, content rationalization, and governance frameworks. While less visible than AI tools, they are foundational. With them, AI becomes trustworthy. There is also a critical human dimension. Legal organizations have traditionally been built on individual expertise. AI shifts this dynamic, creating value from shared, well-organized knowledge rather than isolated experience. This requires cultural alignment, supported by leadership and incentives that promote knowledge sharing. Without it, even the best systems fall short. The legal industry is at a defining moment. AI adoption continues to accelerate, and experimentation is widespread. It is easy to focus on tools. However, tools alone do not determine outcomes. Structure does, clarity does, knowledge does. AI capabilities will continue to evolve, but one principle will remain constant: results ultimately depend on the quality of the knowledge that underpins them. Rethinking KM as strategic infrastructure is no longer optional. The question is no longer whether to adopt AI, but whether the organization’s knowledge is ready for it. Those that address this foundation will not only move faster. They will operate with greater coherence, reliability, and ultimately, greater impact. About the Author María Concepción Stahl, Co-Founder at STAHL SWANSON Consulting María Concepción Stahl is an attorney with international law firm experience, graduating cum laude from Universidad Autónoma de San Luis Potosí and holding a Master’s in Corporate and Business Law from Université Paris II Panthéon-Assas. With over a decade of experience, she specializes in dispute resolution, corporate law, intellectual property, and legal innovation. She served as Knowledge Management Manager at Baker McKenzie Mexico, leading strategic initiatives across the firm. She is now Co-Founder at STAHL SWANSON Consulting, advising on knowledge management, innovation, and AI adoption. Recognized for legal transformation, she has received global awards and is a Chief AI Officer and member of AMCID.
THE CHALLENGE OF PROMOTING FOREIGN INVESTMENT IN A DYNAMIC MARKET: THE CHILEAN ELECTRICITY CASE
I. The Pivotal Role of Foreign Investment in Energy Transition For over four decades, Chile has stood as the undisputed reference for institutional stability in Latin America. Following the pioneering privatization of its electricity sector in 1982, the State established a robust and predictable regulatory framework. Lacking deep liquidity within its domestic banking sector, Chile relied on Foreign Direct Investment (FDI) as the primary engine to construct its modern energy infrastructure. This is why one of the main purposes of this institutional design was to promote foreign investment by assuring investors of regulatory certainty and the protection of their capital. This strategy was wildly successful. In recent years, Chilean installed capacity has successfully pivoted toward non-conventional renewable energy (NCRE), transforming Chile into a global benchmark in the energy transition to carbon neutrality and investments in this sector. Special Laws for Non-Conventional Renewable Energy No. 20,257 (2008) and No. 20,698 (2013) established concrete measures to increase NCRE: they defined renewable energies and mandated power generation companies to supply at least a minimum percentage of their energy injections through NCRE, either directly or indirectly, with severe fines for non-compliance. The purpose was to increase the participation of this kind of generation within the energy matrix by achieving an injection of 20% of NCRE by the year 2025. This goal was wildly surpassed. Today, over 60% of the country’s installed capacity comes from clean sources (solar, wind, and hydro), and for the year 2050, the Decarbonization Plan has set the goal of 100% of the energy produced to be from clean sources. BloombergNEF has consistently ranked Chile among the top ten emerging markets for renewable investment. And this was achieved thanks to political will. The bill through which Law No. 20,698 (which increased the injection of NCRE into the energy matrix up to 20%, as described above) was introduced to Congress explicitly recognized that: “Reaching 20% of NCRE in the electric systems implies strong investments and the development of a series of new technologies that need legal certainty to achieve it. Only with a clear legal framework is it possible to achieve it.” However, regulatory stability is not always consistent. In recent years, we have witnessed important regulatory changes affecting the dynamics of international project financing. For market participants, the question is obvious: How to introduce regulatory changes and at the same time maintain legal certainty? II. NCRE and the Structural Friction on the Chilean Electrical System The rapid boom of renewable energy brought structural friction to the Chilean electrical grid. While clean generation soared, residential electricity tariffs failed to decrease. Concurrently, the national transmission system became severely congested. The massive solar portfolios built in the northern Atacama Desert found themselves structurally isolated, unable to transport power to major consumption centers in the south, triggering energy curtailment. Another relevant point of friction has been the rapid development of Pequeños Medios de Generación Distribuida (PMGD)—small-scale distributed generation projects under 9 MW, which connect directly into the distribution system and benefit from priority dispatch rules and a “stabilized price” regime that protects PMGDs from spot market volatility. This regulatory framework was designed to promote the development of small-scale projects near consumption centers by giving certainty about future income. The result was an explosion of foreign investment in this kind of project. Today, PMGDs represent a formidable 10% of Chile’s net installed capacity, of which 77% is solar. However, their rapid expansion stressed the system’s operational and economic functioning. Large-scale centralized generators began fiercely alleging price discrimination and unfair dispatch priority, targeting the specific “stabilized price” regime. Also, the National Electrical Coordinator started to detect negative impacts on the market’s competition. Under the objective of correcting these “systemic distortions,” a regulatory change was set in motion. What began as a technical debate rapidly mutated into a legitimate expectations dilemma for international financiers. III. The Chronology of Legal Uncertainty The first structural blow landed in 2024 with the debate surrounding the Electricity Tariff Stabilization Law. Seeking to subsidize vulnerable households, the government proposed draining future revenues directly from PMGD projects to fund the state subsidy. Although this specific mechanism was ultimately rejected in Congress after intense pushback, the mere threat of revenue skimming paralyzed the sector. Millions of dollars in planned investment vanished during the legislative debate. The regulatory discussion resurfaced in 2025. The Ministry of Energy and the National Energy Commission (CNE) initiated modifications to Supreme Decrees 88 and 125, fundamentally altering the stabilized price formulas and tightening operational coordination rules for distributed assets (which limits their energy injections, thereby affecting their revenue streams). Once again, international banks and infrastructure funds simply pulled out of the Chilean market, freezing credit lines until the administrative uncertainty settled. These shifts recently materialized into binding regulations under the current administration in 2026. The financial fallout has been evident. Since the discussion of the Electricity Tariff Stabilization Law until the recent publication of normative amendments to the PMGD regulatory framework, investments in the sector have been paralyzed. Recently, the Association of Independent Power Producers of Chile (GIE A.G.) estimated that the regulatory changes may put over US$ 6 billion in committed investments at immediate risk. IV. A Market Participant’s Perspective From my perspective as Legal Counsel managing foreign-backed energy assets, this debate over the regulatory framework reflects a brutal corporate reality. During this multi-year normative erosion, I witnessed investors completely halt the development and construction of two major PMGD portfolios valued at over US$ 300 million. And that’s just me. The collateral damage of a frozen investment pipeline has several consequences. It is a rapid chain reaction of contractual liquidations: The immediate termination of Engineering, Procurement, and Construction (EPC) contracts. The collapse of advanced Share Purchase Agreements (SPAs) and active M&A processes. The termination of long-term land lease agreements. The abrupt cancellation of local supply and consulting contracts, forcing domestic vendors to shutter operations. The strict review of existing credit agreements with regard to the allocation of future cash flows and the ability to cover the debt. The distributed
The Model Is Not the Lawyer: Interaction Quality as Legal Infrastructure
Much of the current conversation about legal AI begins with seductive questions: Can it summarize a contract? Can it draft a clause? Can it compare two versions of an agreement? Can it answer a policy question? Can it support intake, triage, research organization, first-pass drafting, obligation extraction, or document review? These are useful questions. They are also insufficient. In our work, the more serious question is not only what the model can generate. It is under what conditions the model’s output is allowed to matter. That distinction is the line between experimentation and reliance. A model may produce fluent language. It may produce a useful summary. It may even produce something that resembles legal reasoning. But the moment that output influences a decision, a negotiation, a filing, a risk assessment, a contract position, an internal policy, or a business action, the organization is no longer dealing only with technology. It is dealing with reliance and reliance, creates exposure. The model will never be the lawyer. It does not hold professional responsibility. It does not understand the client relationship as a lawyer must. It does not protect privilege by duty (or liability). It does not know when a business risk requires escalation unless the workflow around it has been designed to make that visible. It does not carry institutional accountability. It generates output inside the conditions that humans and organizations create around it. legal AI governance cannot stop at procurement approval, tool selection, or a general instruction to “review AI outputs.” Those measures matter, but they are only the outer wall. The real work begins inside the interaction itself. Legal teams must now govern not only the tool, but the conditions under which the tool is questioned, constrained, reviewed, relied upon, documented, and allowed to leave the room. Tool capability is not the operating model, capability is easy to demonstrate; a platform can produce a contract summary in seconds. It can classify documents. It can identify clauses. It can draft an email. It can propose alternative wording. It can create a neat answer to a complicated question.That is impressive. Alas, It is not by itself, governance. A legal function does not create value merely because a tool can produce output. value appears when that output enters a workflow safely, improves a decision, reduces unnecessary friction, preserves accountability, and can survive scrutiny after the urgency has passed. The real questions begin after the demonstration: Who may use the tool? For which matters? With what kind of data? Under what confidentiality conditions? For what categories of work? With what review standard? At what point may the output influence a decision? Who approves that reliance? What must be documented? What happens when the output is incomplete, uncertain, inconsistent, or too confident? These are not obstacles to innovation. They are the operating conditions that allow innovation to become defensible. Without them, AI becomes another fragmented layer of legal technology: useful in isolated moments, inconsistent across teams, difficult to measure, and dangerous when people begin to rely on it without a shared operating model. The tool may be powerful, but power without operating discipline is not maturity. It is velocity without a brake record; Interaction without quality becomes risk. Many organizations still treat interaction with AI as a user skill. They call it prompting. They offer tips. Advise specifics. Provide metrics and rules for context. Teach how to ask follow-up questions and check the output. That is useful at a basic level, but in law, interaction quality is not merely a productivity technique: It is a risk factor. Who frames the question matters. A poorly framed question can hide the real issue. It can turn a legal risk into a drafting exercise. It can ask the model to optimize language without identifying the commercial context, regulatory constraint, jurisdictional sensitivity, evidentiary burden, internal policy, approval threshold, or risk appetite. The answer may be fluent and still miss the point. What context is provided matters. A model cannot evaluate what it has not been given. If the user omits negotiation history, counterparty behavior, internal approval rules, prior concessions, local law concerns, business urgency, or the reason the question matters, the output may be superficially correct and operationally dangerous. Which assumptions are tested matters. Legal work often turns on assumptions: who has authority, which facts are established, what jurisdiction applies, whether an exception has precedent, whether an obligation is enforceable, whether a risk has already been accepted, and whether the current version of the document is actually the operative one. If the interaction does not force assumptions into the open, the output may create confidence where the organization needed verification. Who reviews the output matters. Review is not a ceremonial glance before copying text into a document. Review is where professional judgment re-enters the workflow. It is where the lawyer asks whether the output is accurate, complete, proportionate, confidential, useful, aligned with the client’s interest, and appropriate for the intended use. When reliance is permitted matters. There is a difference between using AI to explore a question, using it to organize information, using it to draft a preliminary version, and allowing its output to shape a final legal position. Mature legal teams should not treat all AI-assisted work as the same. The risk changes when the output moves from internal support to external consequence. These are not “prompting tips.” They are governance questions. The quality of the interaction shapes the quality of the risk. A strong model used through a weak interaction can produce weak legal work. A disciplined professional using a modest tool, within clear constraints and review standards, may produce a more defensible result. The tool matters. But the interaction around the tool may matter more than we are comfortable admitting. The lawyer remains the accountable interface. One of the most dangerous ideas in legal AI is the casual suggestion that the system is becoming a legal peer. It is not. A peer can be responsible, a peer can be disciplined, a peer can understand duties
