AI and the Law Whitepaper: What’s Working, What’s Not and What’s Next

Nick Abrahams, Legal AI Pioneer, Futurist & Adjunct Professor of Law

Artificial intelligence is already reshaping how legal work is delivered, forcing lawyers, law firms and in-house teams to rethink processes, pricing, training, ethics and client value. This whitepaper maps what’s working, what’s not and what’s next for AI in law. It is structured around seven key sections:

1. The AI Yikes Moment for Lawyers: Introduces the “AI Yikes Moment” through a real case study where a client used a custom AI tool to deliver work dramatically faster and cheaper than a top-tier law firm, challenging the economic foundations of traditional practice.
2. How to Choose the AI Tool for Your Practice: Explains how to navigate the crowded legal AI market using The Breakthrough Lawyer AI Pyramid © to distinguish between Universal, Focused and Precision AI, with practical examples from numerous AI apps.
3. Meta-Prompting – A Simpler Way for Lawyers to Use AI: Outlines meta-prompting, a natural method where you speak to the AI in plain language, let it draft the ideal prompt for you, and then use that prompt to get accurate outputs without needing training to becoming a “prompt engineer.”
4. Strategies for Successful Legal AI Projects: Sets out six practical strategies for making AI projects succeed in legal teams, including being problem-led, ensuring leadership buy-in, starting with business services, investing in training, using gamified innovation like AI Challenges and establishing an AI collaboration space.
5. The Seven Golden Rules of AI: Presents seven essential rules for safe, ethical and effective AI use in legal practice, covering confidentiality, hallucinations, transparency, organisational policies, ethics, IP issues and managing bias.
6. How to Train Lawyers in the AI Age: Explores how to replace the disappearing “training-by-repetition” model with deliberate, structured methods such as simulations, bootcamps, skills labs, AI-powered tutoring and flipped classrooms.
7. Why AI Is Not the End of Lawyers: Argues that AI will transform, not eliminate, legal practice by increasing complexity, unlocking unmet demand, fuelling more disputes and reinforcing the value of human judgement, advocacy, trust and digital intelligence (DQ).

The insights in this whitepaper draw on more than three years of research I have conducted as an Adjunct Professor of Law at Bond University, where I have been examining the impact of AI on legal practice and, in particular, how lawyers can use AI to enhance their work, their service to clients and their lives. Please feel free to contact me as I would be delighted to discuss your AI strategy further, and you may also wish to explore the two online Bond University courses I teach: GenAI Productivity Training for Lawyers and The Breakthrough Lawyer Masterclass.

Section 1. The AI Yikes Moment for Lawyers

I describe an AI Yikes Moment as the point at which AI demonstrates it can perform a meaningful legal task more than ten times faster and more than ten times cheaper than human lawyers, without compromising on quality. It is the moment when theoretical disruption becomes operational reality.

Recently, I witnessed a client’s use of AI that fits this definition precisely. It should command the attention of every lawyer, whether in private practice or in-house.

The client is a major investment fund, a sophisticated buyer engaged in frequent, complex acquisitions of a particular type of asset. They traditionally relied on a leading law firm to conduct due diligence. Each engagement was predictable: three to four weeks of concentrated legal work and a bill of approximately $400,000. For decades, exercises like this have represented dependable, high-value revenue for law firms. Many would view them as part of the profession’s structural “rivers of gold”.

But this client took a different path. Instead of commissioning the next due diligence, they invested $50,000 to build a customised AI tool trained specifically on the type of asset they acquire. This was not experimentation or innovation theatre; it was a deliberate strategic decision to internalise a critical legal process.

They have now deployed the tool on two recent transactions.

The results are remarkable. Each AI-led due diligence only cost them $800 in compute costs and was completed in hours not weeks. Most importantly, the client assessed the AI’s accuracy as comparable to the work previously delivered by their external firm.

Two transactions later, they have saved $800,000 and transformed their view of what legal work should cost.

This is not a prediction about the legal sector’s future. It is a demonstration of what is happening right now, inside real organisations with real consequences for lawyers – in-house and in private practice. Generally, legal AI tools have provided incremental efficiency gains. Useful, yes. Transformative, no.

This, however, is different. When a core, high-margin legal service can be replicated internally at a fraction of the price and time, the economic foundations of the profession shift decisively.

For any lawyer still wondering whether AI will materially reshape legal practice, this example offers a simple, unavoidable conclusion:

Yikes!

Section 2. How to Choose the AI Tool for your Practice

AI offers the potential for profound transformation in the legal industry. It can analyse vast amounts of data, draft complex documents, and even purport to predict case outcomes, with a level of sophistication previously unattainable. Many law firms and in-house teams are already reporting that the impact of AI on legal services has been significant – and this is just the beginning.

The belief in AI’s role in the legal industry exists on a spectrum. On one end are the enthusiasts who envision a future dominated by ‘robo-lawyers’, where AI systems take over entire legal functions. For instance, Goldman Sachs announced that AI could perform 44% of the tasks traditionally handled by lawyers, reflecting a bullish view of AI’s capabilities. On the other end are the sceptics who, having witnessed other tech trends such as blockchain rise and fall without significantly affecting their practice, view AI as just another passing fad.

The reality lies somewhere in between. AI should be seen not as a replacement for lawyers, but as an empowering tool – much like an e-bike for legal work. Anyone who has ridden an e-bike knows you still have to push the pedals and control the direction, but when you hit a steep hill the motor quietly kicks in to give you extra power. AI offers the same benefit by taking the strain out of monotonous, repetitive or low-judgement tasks so lawyers can keep moving smoothly and focus their energy where it matters. In this way, AI enhances capability and efficiency without diminishing the human skill, judgement and nuance at the heart of legal practice. It will augment, not replace, the vital role lawyers play in advising clients, resolving disputes and upholding the rule of law.

Navigating the legal AI landscape

We are currently in the midst of what can be termed ‘peak legal AI’. A plethora of vendors are offering AI-enabled solutions designed to enhance various aspects of legal practice. For lawyers, understanding this landscape is crucial to identifying which tools can most effectively meet their needs and improve their workflows.

As part of my research at Bond University, I have explored how law firms and in-house legal teams are adopting AI. This research led me to create The Breakthrough Lawyer AI Pyramid © framework (see image at Attachment A) which helps lawyers evaluate legal AI products and determine whether they would be a good fit for their legal practice.

This framework categorises AI solutions based on the percentage of lawyers in any organisation (in-house or private practice) that are likely to be able to benefit from the use of the solution’s specific functionality.

Universal AI

At the base of the pyramid (100% of lawyers), we find broad-use general productivity AI solutions that a large majority of lawyers can integrate into their daily practices. These tools are not specifically legally trained but offer general productivity enhancements that are relevant across various legal fields. Examples include AI-driven document summarisation/creation and admin assistance. These solutions are accessible and beneficial to almost every lawyer, making them a foundational aspect of the modern legal toolkit.

Many law firms and in-house legal teams are currently trialling Universal AI tools. It is anticipated that within the next 12 to 24 months, Universal AI tools will become commonplace in legal practice.

The key Universal AI tools are the Public AI applications and Microsoft’s Copilot solution. I discuss these in more detail below. I should note that for the purpose of this paper, I am referring to AI solutions from “Multi-Purpose Vendors” such as Harvey, LexisNexis, Thompson Reuters, Legora and vLex/Vincent as Focussed AI tools but, given the rapid expansion of their capabilities, they are fast moving into the Universal AI category for lawyers.

Public AI Applications

Public AI applications such as ChatGPT, Claude, Perplexity, Grok and Gemini are gaining traction among lawyers. They are accessible on a free or US$20/month basis, making them an attractive option for legal professionals seeking to quickly integrate AI into their practices. However, it is crucial for lawyers to exercise caution when using these platforms, particularly concerning the confidentiality of information. As confidentiality is not guaranteed, sensitive or confidential information should not be entered into these apps. Also, AI often hallucinates (gets things wrong) so all AI output must be verified.

Despite these limitations, these Public AI apps can be effective for a variety of purposes including:

• Search & Summarise: Public AI provides tailored answers (and URL links) to queries and can be prompted by voice.
• Legal & Other Research: Public AI can do research quickly. Most Public AI apps now have “Deep Research” capabilities which takes longer to create the response but the responses are more considered, longer and generally more accurate.
• IT Helpdesk: Public AI apps are great for troubleshooting any tech issues you may have.
• Text translation: Public AI can convert handwritten notes into clean, structured text with correct spelling, whether the source is a notebook page, a whiteboard, or even a photo. It can also extract and organise text from images such as PowerPoint slides, screenshots or photographed documents, turning visual content into editable, searchable text.
• Language translation: Public AI translates documents quickly and cheaply, performing at roughly the level of a junior-translator.
• Tone Translation: Public AI rewrites content to simplify complex ideas or soften overly legalistic or aggressive tone while keeping legal meaning intact.
• Brainstorm Legal Arguments: Public AI generates creative arguments and counter-arguments.
• First-Draft Writing: Public AI produces outlines, clauses and first drafts of documents to help overcome writer’s block and accelerate drafting.
• Regulatory Scanning: Public AI summarises regulatory frameworks and can produce comparison tables as a fast reference before checking authoritative sources.
• Social Media: Public AI converts ideas into clean, professional social media posts with correct spelling and grammar.
• Article Drafting: Public AI drafts outlines and even full articles that are clear and coherent, leaving you to add your voice, stories and nuance.
• Image Prompting: Public AI can create precise images for presentations.
• HR Matters: Public AI can write comprehensive job descriptions, create interview questions (or answers) and assist with preparation for performance reviews.

Microsoft Copilot

Microsoft Copilot is a Universal AI application, which complies with Microsoft’s existing privacy, security and compliance commitments. Therefore many organisations feel comfortable to enter confidential information into Copilot. Furthermore, it can be used in existing workflows (using Word, Outlook, Teams, PowerPoint and Excel).

The uses of Copilot are similar to those set out above for the Public AI tools (but can be done with more confidence around confidentiality). Also, Copilot is available in Teams and Outlook and these have some specific use cases such as:

Copilot in Teams

• Meeting Summarisation: Instantly generates detailed summaries of Teams meetings, highlighting key decisions, action items, and unresolved questions.
• Chat Follow-Up Assistance: Suggests and drafts context-aware responses or follow-ups in ongoing Teams conversations.
• Task Extraction: Automatically identifies and assigns tasks discussed during meetings or chats to the appropriate team members.

Copilot in Outlook

• Email Summarisation: Condenses long email threads into clear, concise summaries highlighting key points, decisions and required actions.
• Drafting Assistance: Creates first-draft responses based on the context of the conversation, your writing style and any attached documents.
• Action Extraction: Identifies tasks, deadlines and follow-ups within emails and automatically adds them to your task list or calendar.

Case Study: Australian Government’s AI Productivity Trial with Microsoft Copilot

The Australian Federal Government conducted a six-month trial of Microsoft Copilot across more than 60 public service agencies. Over 7,600 staff participated.

The trial integrated Copilot into familiar platforms like Word, Outlook, and Teams, enabling public servants to test its drafting, summarisation, and task automation capabilities with minimal disruption. The aim was to assess Copilot’s practical benefits while managing risks in a controlled, whole-of-government environment.

The key results were: 86% wanted to continue using Copilot post-trial. Staff reported saving around an hour per day on routine tasks such as note-taking and email drafting. Notably, 69% experienced faster task completion, and 61% observed improved work quality.

More information here.

Also in the Universal AI category are AI copilots for existing enterprise software solutions. For example, companies such as iManage and Salesforce have deployed AI copilots to assist with existing platforms.

Focused AI

The middle tier of the pyramid consists of legally trained AI solutions that address specific legal use cases, such as due diligence or contract review and e-discovery. These use-case-specific legal AI solutions are designed to tackle more specialised tasks and will not be relevant for all lawyers, all the time. Their targeted nature allows for more precise and efficient handling of specific tasks, providing significant value to practitioners in those areas.

There is a large number of Legal AI vendors and I do not propose to analyse them all. Please see The Breakthrough Lawyer AI Pyramid © at Attachment A to get a sense of some of the vendors and their offerings.

Case Study: Harvey’s Rapid Impact on Legal Workflows

Harvey is a legal AI rocket ship. It only started in 2022 and at its most recent capital raise of US$150 Million in October 2025, it was valued at US$8 Billion.

It seems Harvey is really living up to its valuation. Research released in November 2025 shows just how much lawyers are benefitting from the platform.

Research company RSGI has independently surveyed 40 Harvey customers around the world. The feedback? Strong usage, fast impact and a sense that Harvey is becoming essential legal tech.

The most exciting impact (for me at least) was on lawyers themselves. They reported being able to spend more time on meaningful, strategic work and less time on drudgery. Longer Harvey usage correlated directly with higher workplace fulfilment

Key Findings from the Harvey Impact Study

• 100% of law firms agreed their lawyers would be upset if Harvey access were removed, and 90% said Harvey improved workplace fulfilment.

• Firms strongly agreed Harvey helped them reduce non-billable work (93%), strengthen client relationships (83%), and deliver work faster (80%).

• In-house teams were more inclined to see time savings and reported delivering work faster to business colleagues. 90% said Harvey increased their team’s overall capacity.

• Power users – typically 20–30% of users – were seen as the biggest drivers of adoption, usage and value. Power users saved an average of 34.1 hours per month, more than double the standard user average of 14.5 hours.

• Usage is high: 92% of licences are used monthly. 68% of organisations saw value within three months, and one-third saw benefits within the first month.

• 55% of participants reported that Harvey users performed better than non-users; others said it was too early to quantify.

• Harvey also opened doors to more client face time, enabling discussions about new fee models, product innovation and deeper strategic collaboration.

• Around half saw clear cost-savings or new revenue products – with a lot of the others saying they were “on the cusp” and viewed Harvey as essential infrastructure for the future.

The study was commissioned by Harvey but conducted independently by RSGI. They interviewed 40 organisations including 29 law firms and 11 corporate legal departments across North America, Europe, the UK, APAC and Latin America, representing organisations of all sizes and levels of AI maturity.

While it is still too early to say what the future is for legal AI, these early metrics are impressive.

More information here.

Precision AI

At the peak of the pyramid are AI solutions aimed at assisting lawyers by providing 100% accurate legal analysis of complex legal issues – behaving almost like a super-lawyer. These tools will be used less often by lawyers as most lawyers do not spend a lot of time advising on extremely complex legal issues. These solutions will represent the cutting edge of what AI can achieve in the legal domain.

However, it is very difficult for the machine alone to reach the level of accuracy required by lawyers. Consequently, the outputs of these solutions always need to be reviewed by the legal professional.

This area is being primarily targeted by Multi-purpose Legal AI vendors who are endeavouring to leverage advanced AI capabilities to provide nuanced insights and analysis.

Section 3. Meta-Prompting: A Simpler, More Natural Way for Lawyers to Use AI

There has been plenty of commentary suggesting that lawyers must become “prompt engineers” to use AI effectively. The message often implies that mastering AI requires learning complex prompt structures, frameworks and formulas. This has created anxiety for some legal professionals. The expectation becomes a barrier to adoption: “What if I don’t phrase it correctly?” or “What if I don’t know the right structure?”

In reality, most lawyers do not need to study the technical art of prompt engineering. A far more accessible approach, and the one I recommend, is called meta-prompting.

Meta-prompting simply means telling the AI everything it needs to know so that it can write the perfect prompt for you.

The easiest way to do this is to turn on your AI app’s microphone and speak in a natural, stream-of-consciousness way. You do not need to worry about grammar, order, structure or eloquence. Just talk through what you are trying to achieve, the context, the constraints and the outcome you need. Speaking rather than typing makes the process fast and intuitive and removes the fear of “getting it wrong.”

Once you have finished explaining, you ask the AI:

“Please draft the prompt I should give you to perform this task.”

The AI will generate a clear, structured and detailed prompt based entirely on your spoken input. You simply review it to confirm it reflects what you want and then ask the AI to run it.

The process feels counter intuitive. It is like asking a famous chef to write the recipe for their signature dish and then handing the recipe straight back and asking them to cook it. But it works remarkably well. Meta-prompting lowers the barrier to entry and allows lawyers to use AI with confidence without needing to become prompt engineers.

Section 4. Strategies for Successful Legal AI Projects

Strategy 1: Be Problem-Led, Not Product-Led

A key lesson from the past few years of legal AI adoption is that many legal teams have invested heavily in what can be described as “product-led” innovation. This occurs when teams purchase legal AI tools without a clearly defined problem or use case. The expectation then falls on already time-poor lawyers to experiment with the new tool and identify potential applications that might deliver value. In practice, this model rarely succeeds. Lawyers often do not have the spare capacity required to trial emerging technologies, and the absence of a defined problem means that experimentation can be unfocused and slow. As a result, many product-led initiatives fail to deliver meaningful impact.

A more effective approach is problem-led innovation. This is how innovation typically happens in successful organisations, including start-ups. A problem is first identified, clearly articulated and prioritised. Only then is a project launched to address that problem through a combination of technology and process improvement. The goal is not to “use AI” for its own sake, but to solve a real pain point in a measurable way.

For legal teams adopting AI, the same principle applies. Begin with a well-defined problem. Select AI tools that directly support the solution. And test against clear metrics so you can determine whether the AI is actually delivering improvement.

Strategy 2: Leaders Must Be Visible, Active Champions of AI

Successful AI adoption in legal teams depends heavily on visible, consistent leadership. Lawyers need to see that engaging with AI is not only permitted but actively encouraged by those at the top. When leaders model AI use in their own work and signal its strategic importance, uptake accelerates and hesitation fades.

Across many large organisations, it is now the senior executive team and even the board driving momentum. They want their organisations to have a credible AI story and are personally adopting AI tools to improve decision-making and responsiveness. A striking example comes from Peter Tonagh, Chair of stock market-listed company, GTN. Tonagh received a takeover bid for GTN at 9:30 a.m., just 30 minutes before the stock exchanged opened. Instead of first calling the usual advisers – the investment bank and external law firm – he ran the bid documents through his AI app. Within moments, the AI produced a step-by-step action plan for how he and GTN should respond. It also drafted a market announcement and press release. Only after reviewing the AI’s work did Tonagh contact external advisers. The guidance he received from the AI proved accurate and immediately actionable.

This example shows how modern leaders are not waiting for AI to trickle upwards. They are using it themselves, setting the tone and demonstrating that AI is a powerful tool for judgment-heavy, time-critical decisions. Legal team leaders should do the same: lead from the front, use AI openly and signal that the organisation expects and supports responsible experimentation.

Strategy 3: For Law Firms, Start in the Business Services Teams

For many law firms, the fastest and most reliable return on investment from AI comes not from legal work, but from the business services teams that keep the firm running. Functions such as marketing, finance and HR often contain high-volume, document-heavy, repeatable tasks that are ideally suited to AI support. These areas frequently represent the true “low-hanging fruit” for early AI deployment and can demonstrate measurable value far more quickly than lawyer-facing use cases.

Firms should therefore prioritise giving marketing, finance and HR teams access to AI tools before rolling them out widely to lawyers. In marketing, AI can significantly speed up the creation of proposals, pitch decks, capability statements and responses to RFPs. These tasks are time-sensitive, template-driven and often non-billable, making them an ideal early proving ground for AI.

A finance team in a law firm could use AI to automate billing reviews, streamline trust accounting checks and generate real-time financial insights that improve forecasting and decision-making.

In HR, AI can streamline the development of job descriptions, support performance review processes, assist with policy drafting and help manage internal communications. These improvements can reduce administrative burden, increase accuracy and free HR professionals to focus on higher-value initiatives.

Strategy 4: Invest in Training and Change Management, Not Just Technology

Many legal teams have already invested in powerful AI platforms such as Copilot, CoCounsel, Harvey and Lexis+. These tools are extraordinarily capable, yet my research shows that access alone does not translate into meaningful adoption. Lawyers often find it difficult to identify practical, high-value applications on their own. With demanding budgets and heavy workloads, they simply do not have the time required for extensive experimentation. The result is predictable: sophisticated AI sits under-used, and the expected return on investment fails to materialise.

Recent research strongly supports this finding. A 2025 MIT study revealed that 95% of enterprise AI pilots fail, not because the underlying technology is inadequate, but because of what MIT calls the “learning gap”. Organisations buy the tools, but their people do not know how to use them effectively, and the tools themselves are rarely configured to reflect real workflows.

Similarly, 2025 research from Boston Consulting Group (BCG) found that organisations consistently overspend on technology and underspend on training and change-management. BCG’s analysis shows that successful AI programs follow a clear budgeting pattern: 10% on the models themselves, 20% on the data and integration, and 70% on changing the way people work, especially through targeted training.

For legal teams, this means that purchasing AI is only the opening move. The firms that succeed in an AI-enabled future will be those that equip their lawyers to know what to delegate to AI, what to keep close, and how to apply sound judgement when AI is doing the heavy lifting.

Strategy 5: Harness Gamified Innovation to Accelerate Adoption

One of the most effective ways to build practical AI capability within legal teams is through structured, gamified innovation, such as an AI Challenge. These initiatives create a focused, time-bound environment in which lawyers can experiment with AI, learn collaboratively and develop solutions that address real operational needs. Gamification reduces the pressure often associated with traditional innovation programs and replaces it with curiosity, engagement and measurable outcomes.

A good example of this approach is Norton Rose Fulbright’s Global AI Chatbot Challenge, which I led while serving as the firm’s Global Head of Technology & Innovation. The challenge enabled lawyers, regardless of technical ability, to build client-facing AI chatbots using a no-code platform. It produced four fully operational Parker bots across Australia, the UK, South Africa and Europe, covering privacy and insurance law. Despite costing less than $1,000 in total per bot to build and operate, the Parker bots generated significant amounts of client work.

Another example is Singapore General Hospital’s Chatbot-a-thon, an AI innovation challenge that brought together over 150 staff from 22 departments. Participants created 48 AI assistants using a no-code platform, five of which have since moved into daily operational use. The initiative successfully empowered non-technical clinicians to design tools that improved workflow efficiency, clinical support and patient care.

While law firm Reed Smith has a scheme that lets lawyers count up to 50 “Innovation Hours” a year toward their billable target, encouraging them to design new tools, processes and client-facing products. Junior and senior lawyers alike will spend this time scoping problems, prototyping workflow improvements, and collaborating with the innovation team to develop practical solutions. A pilot last year saw 17 lawyers contribute 364 hours to six projects, chosen from 30 ideas submitted across the firm based on creativity, client value and strategic impact. One of the selected outcomes was the Breach RespondeRS app, built entirely in-house to simplify US data breach laws. Reed Smith will now fast-track five new projects annually, embedding innovation time as a formal part of legal work rather than an extracurricular effort.

On a personal level, I am a big believer in the power of such challenges and have worked with the legal teams at Pfizer, ANZ Bank and the NSW Government to implement gamified innovation programs and have consistently found that lawyers respond positively to the format and develop practical, high-value solutions.

Strategy 6: The First Rule of AI Club is to Talk about AI Club

A powerful strategy for successful AI adoption is to create an “AI Club” within your organisation. This is an online space where people can openly share how they are using AI in their day-to-day work. Many employees experiment with AI quietly, worried about how they might be perceived, but a culture of openness and transparency removes that hesitation and encourages broader, more confident use.

By giving people a simple collaboration platform to post examples, tips and successes, you unlock peer-to-peer learning, which is one of the most effective drivers of behavioural change. And, of course, it’s important to remember the first rule of AI Club: You must talk about AI Club.

Section 5. The Seven Golden Rules of AI

Rule 1 – Do Not Enter Confidential Information into Public AI Apps

Confidential information should never be entered into ChatGPT or similar public AI systems. These platforms process data through external servers and may store information in ways that cannot be fully controlled or audited by your organisation. While vendors emphasise privacy protections, no provider can guarantee absolute confidentiality or prevent future policy changes, security vulnerabilities or unintended data exposure. The safest approach is to assume that anything entered into a Public AI system could, in theory, be seen by others.

Rule 2 – Remain Vigilant About Hallucinations

Generative AI models are highly capable, but they often produce incorrect or fabricated information with substantial confidence. These “hallucinations” can include invented cases, inaccurate facts or entirely fictional citations. For lawyers, this presents an obvious risk. Always verify outputs against authoritative sources and never rely on AI-generated content without independent checking. Think of AI as a junior lawyer with exceptional speed but inconsistent accuracy. It can dramatically improve your efficiency, but it still requires supervision. Maintaining a disciplined review process ensures you benefit from AI’s strengths without compromising the quality or integrity of your work.

Rule 3 – Be Transparent About AI Use

Transparency about your use of AI builds trust with clients, colleagues and other stakeholders. Letting others know that you have used AI in your workflow demonstrates professionalism and signals that you are using modern tools responsibly. A simple approach is to explain that AI assists with efficiency, but the analysis, judgement and final review remain human-led. Clear communication ensures others understand the role AI played and reinforces that accountability always stays with the lawyer.

Rule 4 – Follow Your Organisation’s AI Policy

Lawyers must comply with their organisation’s policies governing the use of AI. If your firm restricts or prohibits certain tools, those rules must be respected. However, it is important to note that in many organisations, employees are using AI tools despite policies that formally limit or forbid their use. This often reflects a gap between policy and workplace reality.

If you are working in an environment where AI use is tightly restricted, it may be appropriate to discuss this with management and explore whether the policy can be updated to provide a more practical and safer framework for AI adoption. For example, the New South Wales Government permits the use of ChatGPT, but only in accordance with detailed guidelines. This type of structured, risk-aware approach may be a useful model for organisations seeking to modernise their AI policies while maintaining appropriate safeguards.

Rule 5 – Maintain Ethical and Professional Responsibility

Your use of AI must always align with the ethical rules of your jurisdiction. Some courts restrict or require disclosure of AI assistance in filings, so litigation practitioners must remain updated on court-specific requirements. More broadly, professional responsibility obligations continue to apply: lawyers must maintain competence, ensure accuracy and act in the client’s best interests.

Importantly, technology can support ethical practice. In some jurisdictions, lawyers have an obligation to keep fees reasonable, and responsible use of AI may in the future help reduce costs.

Rule 6 – Intellectual Property Considerations

One of the benefits of generative AI is that it produces new text each time it generates an output. You can ask ChatGPT the same question three times and you will receive three different responses. While the themes and underlying ideas may be consistent, the wording will vary. This means the risk of reproducing copyrighted text is very low, as the system is not retrieving or repeating existing material but generating fresh content. The same principle applies to AI-generated images, which are created anew for each request.

However, as image-generation tools become increasingly powerful, they are now capable of producing visuals that closely resemble real individuals, commercial logos or trademarked designs. For this reason, lawyers should exercise particular caution with AI-generated images. Always review outputs carefully to ensure that they do not unintentionally replicate protected material or depict identifiable individuals without proper rights or consent.

Rule 7 – Be Aware of Bias and Promote Inclusive Outputs

AI systems learn from vast datasets, much of which reflects the biases present on the internet. As a result, AI outputs can unintentionally reproduce stereotypes or underrepresent certain groups. Lawyers should therefore be intentional when crafting prompts, especially for images, by requesting diversity and inclusive representation. Being conscious of potential bias helps ensure that AI-generated content aligns with professional standards, organisational values and broader social expectations.

Section 6. How to Train Lawyers in the AI Age

For decades, the legal profession relied on a predictable training pipeline. Junior lawyers learned by doing the repetitive, labour-intensive work that forms the base of many legal matters: due diligence, discovery review, document comparison, basic research and first-draft preparation. This work served a dual purpose. It offered juniors a structured way to build foundational skills through repetition, and it allowed law firms to bill clients for the juniors’ time while they learned. It was, in many ways, an elegant economic model: juniors gained expertise while firms captured revenue.

AI has now disrupted this long-standing apprenticeship. AI systems can already review documents, analyse contracts, flag issues, synthesise case law and generate first drafts. As AI capability grows, these once-reliable training grounds are disappearing. Work that previously helped a junior lawyer “learn the ropes” is increasingly automated, inexpensive and unbillable. This is positive for clients and firm efficiency, but it creates a significant challenge: how do we train junior lawyers when the work that traditionally trained them no longer exists?

Lessons learned from other professions

We will have to change how we train junior lawyers. We can learn from other professions that have already been through this challenge. In the medical field, they used to train doctors through exhausting on-call work and repetitive ward tasks. Safety scandals and work-hour limits forced a shift to simulation labs and structured curricula, which have improved patient outcomes and reduced errors. Nursing moved from pure ward-based learning to a blended model after evidence showed high-fidelity simulations and class-based training could safely replace a significant amount of the clinical hours, producing graduates who were just as competent and often more confident.

Aviation once relied on new pilots learning by flying real aircraft and slowly graduating from junior cockpit roles. Catastrophic crashes drove the introduction of full-motion simulators and crew resource management, helping make commercial flight extraordinarily safe. Accounting firms historically trained juniors through manual low-level audit work, but automation and AI pushed them to create intensive bootcamps and AI-supervision training, with some firms reporting that first-years now reach “fourth-year” capability much faster.

Training in engineering and architecture once centred on hand-drafting and manual calculations, but the rise of technologies such as CAD reduced the need for those traditional skills. Today’s junior architects and engineers are developing their skills through redesigned university programs, immersive simulations and structured digital training pathways, leading to more consistent competence across the profession.

Across all these fields, when traditional entry-level work disappeared, they replaced it with intentional, often simulation-rich training. Done well, the outcomes were as good or better.

New training models

So the central question is not whether lawyers can be trained without the traditional entry-level tasks of the profession. They can. The real question is who will pay for the training, and what methods will deliver the best results at reasonable cost and speed. The answer will vary by firm size and jurisdiction, but the training models emerging globally provide a clear roadmap.

Below are the most viable approaches to training junior lawyers in the AI era, along with real-world examples demonstrating each model in action.

1. Simulation-Based Training

Some firms are now training junior lawyers the same way airlines train pilots – through high-fidelity simulations. Instead of waiting years to “see enough deals,” juniors are thrown into realistic, end-to-end scenarios with mock negotiations, complex transactions and time-pressured decisions. All with guidance from senior practitioners. It’s an efficient way to build judgement and pattern-recognition quickly.

DLA Piper’s Advanced Negotiations and Structuring Academy (ANSAi), is a “legal flight simulator” that blends AI-powered scenarios with expert-led mentoring to build the high-level judgement and negotiation skills juniors can no longer learn through routine work. Over six weeks, lawyers act as principals in a complex, shifting deal environment, supported by human experts, while AI dynamically introduces ambiguity, conflict and real-world pressure.

The firm sees this as essential preparation for a future where AI removes much of the traditional junior workflow, leaving lawyers responsible for higher-value strategic thinking that machines cannot replicate. While not every lawyer will train this way, DLA believes AI-integrated experiential learning will become a major differentiator for firms that want to attract talent and deliver strategic value to clients.

Rubi Legal Training is a new virtual apprenticeship platform created by former BigLaw M&A lawyers to close the gap between legal education and real-world transactional skills. Its flagship M&A program simulates the experience of a junior associate on a complex deal, guiding participants through tasks, documents and judgement calls. The University of Texas School of Law has adopted Rubi for its entire student body, recognising its value in preparing students for transactional practice. The founders plan to expand the platform into additional practice areas and seniority levels, building a broader training ecosystem.

Kennedys has partnered with legal AI solution, Spellbook to create a training program designed to rebuild learning opportunities lost in an AI-driven world. The program combines Kennedys’ legal expertise with Spellbook’s drafting and analysis tools to teach judgement, reasoning and AI-assisted practice. Junior lawyers work through simulated matters and AI-supported drafting exercises, receiving structured feedback that mirrors the coaching once gained through live files. Pilots are happening in the UK and US offices.

2. Bootcamps and Structured Onboarding Programs

Some firms are taking a page from the tech world and running multi-week “bootcamps” for new lawyers before they touch any client work. These programs blend teaching, drafting drills, simulated matters and practical exercises so that juniors start their practice already equipped for higher-level tasks.

Goodwin’s award-winning first-year associate program blends internal expertise with external learning tools to deliver comprehensive training suited to the AI era. Over eight weeks, associates work through immersive business and litigation simulations, learning foundational concepts through external video and practice platforms before engaging in live, partner-led sessions. This structure preserves valuable partner time for firm-specific insights while ensuring scalable, consistent skill-building across the cohort. The result is earlier development of judgement, confidence and an ownership mindset – capabilities that will help associates use AI tools effectively and meet evolving client expectations.

3. Business & Professional Skills Training

I was fortunate over decades in practice to learn business and professional skills directly from exceptional partners – simply by working alongside them. I never had formal mentoring, but the chance to observe seasoned practitioners and discuss real matters with them provided invaluable training that shaped my career. Those opportunities have diminished sharply in recent years as time pressures, digital drafting and virtual communication have reduced the personal contact through which those skills were once passed on almost osmotically. The Breakthrough Lawyer Masterclass gives lawyers structured access to the kind of practical, experience-based learning that once occurred naturally.

The Breakthrough Lawyer Masterclass is a live-virtual program I deliver through Bond University to help lawyers access the high-quality training and mentorship that modern practice often no longer provides. Limited to 40 participants per cohort, it has already supported 75 lawyers globally. Over six weeks, lawyers learn practical, future-focused business, professional and personal skills through live-virtual sessions, self-directed online learning, one-on-one coaching and insights from more than 30 global general counsels and business leaders.

4. Gamified Innovation

One effective way to build practical AI capability in legal teams is through the structured, gamified innovation programs that I mentioned in Section 4 above. Good examples are Norton Rose Fulbright’s Global AI Chatbot Challenge, Singapore General Hospital’s Chatbot-a-thon and Reed Smith allowing lawyers to count 50 “Innovation Hours” towards their budget.

5. AI-Powered Tutoring and Real-Time Feedback

AI is not just changing legal work – it is also becoming a powerful training tool. Firms are increasingly using AI writing coaches, clause-analysis tools and feedback engines to give juniors instant critique on research, writing and drafting. This dramatically compresses the learning cycle.

Wilson Sonsini recognised that associates were struggling to assess automated outputs from its incorporation platform, Neuron, because they lacked a deep grounding in market terms and foundational concepts. In response, the firm built a comprehensive training program combining pre-learning, small-group mentoring and hands-on exercises, including drafting incorporation documents manually before using automation. This experiential, multi-phase approach strengthens judgement, improves issue-spotting and restores the kind of practical learning traditionally gained through close partner interaction. The program has been so effective that the firm now plans to expand similar training into other practice areas.

6. Flipped Classroom and Online Learning Models

Another trend is “flipped” learning. Instead of partners giving long lectures, juniors first complete short online modules at their own pace, covering legal concepts, industry basics or AI literacy, and then attend live sessions that focus entirely on application and problem-solving.

Crowell & Moring uses video-based learning to give lawyers on-demand access to essential concepts while reserving live sessions for deeper discussion and practical application. This approach underpins major initiatives such as their mini-MBA and AI training, where curated content builds foundational understanding before lawyers attend live workshops.

The strategy allows the firm to tailor content to different audiences, increase engagement, and make in-person time far more effective because participants arrive prepared with questions. Overall, the blend of video and live learning delivers flexible, comprehensive training that supports lawyers at all levels while reducing the burden on the professional development team.

7. Manage AI Agents

KPMG is now training its newest consultants to manage teams of AI agents – digital assistants capable of completing complex tasks autonomously. The aim is to shift juniors from low-level work to strategic thinking, using AI-generated insights to contribute meaningfully in high-level discussions. While the approach is still being tested, KPMG expects future consultants to work alongside curated teams of AI agents that expand their capability and accelerate their development.

Conclusion

AI will remove the repetitive work that trained junior lawyers for generations. But it hasn’t removed the skills lawyers need. Those skills simply need to be taught differently, more intentionally, more creatively and far more efficiently. The legal teams that adopt structured, experiential approaches like simulations, bootcamps and AI-assisted coaching will develop junior lawyers faster, better and with far greater confidence than the old apprenticeship model ever allowed.

Section 7. Why AI is not the end of lawyers

Famed legal commentator Professor Richard Susskind recently began a Times article with a dire forecast: ‘If the leaders of most artificial intelligence companies are right about the speed of technological advance, there will be little work left for traditional lawyers by 2035.’

Susskind suggested that machines would take over from lawyers in his 2008 book The End of Lawyers? Nearly 17 years later, there are more lawyers than ever.

The recent surge in AI development has sparked concerns about the future of many professions, law included. Some, like Susskind, predict that machine intelligence will massively displace lawyers. However, there are compelling reasons to believe that the future for lawyers remains bright, albeit different. If anything, AI will create new opportunities rather than wipe lawyers out.

Technology creates complexity

Every technological advance has legal consequences. Far from reducing legal work, technology often creates more of it. Take email, for example. It revolutionised communication, but it also expanded the scope of legal discovery exponentially. Pre-email, discovery was largely limited to letters and memos. Today, lawyers can be tasked with reviewing thousands, sometimes millions, of emails in a single matter. Now imagine the scale of discovery when full AI-generated transcripts of every meeting are routinely created and retained.

AI will make the world a more complex place, and lawyers thrive on complexity. Privacy law was not really a thing 10 years ago, but the rise of technology has made privacy a major legal issue. Many lawyers now do privacy work. The same with cybersecurity. So it will be with AI.

Capability does not equal adoption

The mere capability of a technology does not guarantee its adoption. Historically, there is often a long lag between invention and widespread use. Many factors – including cost, complexity, and resistance to change – slow down the uptake of even the most promising innovations. The law is no exception. Tech deployments are expensive and often difficult, regulatory issues are thorny and entrenched practices die hard. AI will certainly reshape legal work, but it will not do so quickly.

Meeting unmet legal demand

There is a vast, unmet demand for legal services. Traditional legal advice remains too expensive for many individuals and small businesses. AI will allow lawyers to offer services at a much lower cost, unlocking this under-served market. This phenomenon is explained by the economic theory known as Jevons Paradox. The paradox is that as technology improves efficiency and reduces cost, overall demand often increases rather than staying constant. Think of your phone: the cost per megabyte of data has dropped dramatically in recent years, but your phone bills have not. Instead, we find more things to do with our phones – streaming, scrolling, downloading – and end up using more data. The same dynamic applies to legal services. As AI makes legal advice cheaper and more accessible, more people will seek it out.

The company I co-founded, Lawpath, demonstrates this potential. Lawpath has served more than 500,000 clients – many of whom would likely have gone without any legal assistance if faced with traditional law firm fees. Rather than eroding legal work, AI can expand the market, allowing lawyers to meet previously unmet legal needs.

More disputes, not fewer

AI is also likely to fuel a significant rise in the number of disputes. Recently, the US Treasury issued a new crypto rule and expected around 200 public submissions in response. However, a group of crypto advocates created an AI app that allowed users to generate submissions at scale, and the agency received over 120,000 submissions – which they had to review.

Likewise, many companies are reporting a spike in complex customer complaints, thanks to AI tools such as ChatGPT making it easier for consumers to lodge grievances. The Harvard Business Review reported that ‘making a complaint’ ranked number 23 out of the top 100 uses for AI. Recently UK regulators approved a new AI-powered law firm, Garfield AI, that will do letters of demand for £2.

All of those grievances need to be reviewed and settled – more work for lawyers on the receiving end.

We are seeing much investment in technology to empower the plaintiff bar – those firms that bring mass tort cases. For example, AI platform Eve recently raised $47m to ‘help plaintiff law firms punch above their weight’. AI-powered legal analysis platform Supio raised $60m to help personal injury and mass tort firms review medical bills and deal with all aspects of these cases.

The most innovative of the developments in the mass tort space is ClaimClam. Just enter your details, and it will show you which class actions you are eligible to join.

AI democratises the legal process, and the judiciary is preparing itself for a barrage of self-represented litigants. A New York State Supreme Court judge recently admonished a self-represented litigant for using an AI-generated avatar to plead his case for him.

AI will likely make the world a more contentious place – and where there is conflict, there is demand for lawyers.

Human nature still favours lawyers

Human nature plays a crucial role in preserving the need for lawyers in three important ways.

1. People want to win

People could resolve disputes or complete deals earlier, but often choose not to. Even when the facts and the law are clear, human psychology comes into play. People want just that little bit more; to feel they have won. If parties cannot agree, it does not matter what a machine says: lawyers are needed to bridge the gap and broker a settlement.

2. People level up, not give up

People do not surrender to technology; they level up. Geoffrey Hinton, the Nobel prize-winning ‘godfather of AI’, famously predicted in 2016 that AI would render radiologists obsolete within five years. Yet the number of radiologists in the US has grown by 1.4% annually since then, and there is a global shortage of radiologists. Rather than abandoning their careers and becoming baristas, radiologists adapted. They now use AI tools to provide a better service to patients.

In the 1980s, it was said that Microsoft Excel would reduce the number of accountants. But according to Morgan Stanley, the number of accountants has increased significantly since Excel’s introduction. Accountants use the tool to provide greater insights to clients.

Lawyers can, and must, do the same: adopting AI to enhance, not replace, their practice.

3. People seek human expertise

In a world filled with dizzying amounts of information and disinformation, people gravitate towards human experts. We have been able to book travel online for decades, yet travel agents are still thriving. At Lawpath, we initially struggled to sell a completely DIY, ‘ask the machine’ legal service. There was little interest in a fully DIY solution. The success of the business came from offering a blend of machine efficiency and human lawyer advice and oversight. Clients still want reassurance from a trusted expert, especially in areas they do not understand, like law.

A call to adapt

None of this means that lawyers can afford to be complacent. The profession will change. Lawyers will need to develop new skills, both technical skills to work with AI and human skills to complement it. Communication, creativity, empathy and judgement will become even more critical. AI can deliver information; only humans can deliver understanding, trust and advocacy.

The future of law is not human versus machine; it is human with machine. Lawyers who embrace AI as a tool to do better work, faster and more affordably, will find themselves in greater demand than ever. The skills that have always been at the heart of great lawyering – critical thinking, problem-solving, persuasion – will not disappear. They will be enhanced.

Lawyers have long relied on IQ and EQ but, as AI improves, they will now need to cultivate DQ – their digital intelligence. DQ is the ability to understand, relate to and collaborate with almost-human-level AI.

The legal profession has survived many incredible advances in technology. Each wave of innovation has changed the practice of law but also expanded its reach and relevance. AI will be no different. The lawyers who succeed will be those willing to adapt, learn and lead – not fear the future, but shape it.

Nick Abrahams

Please feel free to contact me

• Nick@NickAbrahams.com
• Nick on LinkedIn

Background on Nick Abrahams

Nick Abrahams – Legal AI Pioneer & Futurist

Nick doesn’t just talk about the future; he’s building it. He created the award-winning Parker, the world’s first AI-enabled privacy chatbot. Parker was rolled out in four countries and generated more than $1 Million in legal fees. He is a Co-founder of Lawpath, an AI-powered legal services platform with over 500,000 clients. Lawpath recently received an investment of $10 Million from a major bank. Nick was a partner at international law firm, Norton Rose Fulbright for twenty-five years, spending the last ten years as a global technology and innovation leader.

His hands-on work in AI earned him an appointment as an Adjunct Professor of Law at Bond University, where he researches how AI can enhance the legal profession. He shares his findings through two popular courses in The Breakthrough Lawyer Program: GenAI Productivity for Lawyers and a high-performance Lawyer’s Masterclass.

Nick is the winner of the Financial Times newspaper’s Asia Innovator of the Year Award and has advised both Berkeley and Stanford Universities on their Legal AI training programs. He works extensively with organisations and governments on Legal AI strategy, training, implementation and governance – including recent collaborations with the legal teams at the United Nations and at the International Monetary Fund.

He is a LinkedIn Top Voice in Technology, the author of two Amazon bestselling books, Big Data, Big Responsibilities and Digital Disruption. He is on the boards of the Vodafone Foundation and, world-leading genomics research organisation, the Garvan Foundation. He is the host of the Web3 Goes Mainstream podcast.

Before his legal career, Nick graduated from the University of Southern California Film School and worked as a Creative Executive at Warner Bros. in Los Angeles on the hit TV shows, ER and The West Wing. He is a former professional stand-up comic and is the founder of The Tokyo Comedy Store. More information on Nick here.