What is legal tech?
In a broad sense, «legal tech» («tech» for technology) stands for the digitalisation of the legal sector, i.e. law firms, corporate legal departments and the judiciary. This includes the digital preparation of legal advice and legal processes by the clients themselves, the digitalisation of communication, e.g. through online advice, the automated creation of legal documents such as contracts, even self-executing contracts, so-called smart contracts, digital tools for document analysis, e.g. in the context of due diligence, artificial intelligence (AI) to support lawyers and judges, e.g. to calculate the chances and risks of litigation, and the complete digitalisation of the justice system, right through to automated court proceedings. However, AI that renders judgments itself is generally rejected: the final decision must lie with judges.
The world’s first internet court
In 2017, the world’s first internet court opened in the Chinese city of Hangzhou, home of the IT giant Alibaba Group. It deals with disputes arising from online activities, entirely online. Beijing and Guangzhou followed in 2018.
Sources: Hangzhou Internet Court (official website, English); video CGTN (Chinese state broadcaster); LTO 29.09.2017 (German, use the translation function); China Justice Observer.
Until recently, a distinction was made between digital tools that are also used by the legal sector, such as word-processing programs, and tools developed specifically for the legal sector; only the latter counted as legal tech. Generative AI has blurred this line. Claude, ChatGPT & Co. were not built for lawyers, yet they now perform core legal work: analysing cases, assessing legal questions, drafting contracts and briefs, always within the limits of professional secrecy, which means that lawyers may not feed confidential client information into tools that do not guarantee its protection. Specialised legal AI products such as Harvey or Legora are themselves built on these general-purpose models. Today, legal tech is therefore better defined by its use than by its origin: any digital tool applied to legal work.
With legal tech, traditional providers of legal services, such as law firms and trust companies, face competition from legal tech companies that offer their services exclusively online. Advice on flight delays, parking fines or administrative orders can be obtained simply, quickly and cheaply from online services such as Flightright or DoNotPay. Their business model is based on the automation of standard cases. An overview of legal tech applications can be found on digilaw.ch, chapter 14.03 (German; use the translation function).
Case: DoNotPay
DoNotPay is a US online service, founded in 2015, that helps consumers contest parking tickets, cancel subscriptions or claim refunds with automated letters and forms. It advertised itself as «the world’s first robot lawyer». In 2024 the US Federal Trade Commission took action: the company had not tested whether its output matched the quality of a human lawyer and had no lawyers on staff. In February 2025 DoNotPay accepted an order to pay USD 193,000 and to stop claiming that its service can replace a lawyer without evidence.
Lesson: legal tech is subject to the law like any other business, in particular to the rules against misleading advertising.
Source: FTC press release, February 2025.
Legal tech for consumers
Ask a chatbot of your choice (Claude, ChatGPT & Co.): «Which online legal services can I use as a consumer in [my country], for instance for flight compensation, parking fines or contract templates? Which of them are international, which are local? Give me the sources.»
Validate the answer by opening the providers’ websites and checking whether they actually serve your country.
In class: in pairs, compare which services you can use and what they cost.
What this means for you: As a manager, founder or employee you will use AI as the first port of call for legal questions: to understand a contract, assess a risk or find out how to proceed. This is faster and cheaper than ever. The limit: when the matter is important or you are unsure, consult a lawyer, with the groundwork already done. Legal tech does not replace lawyers; it makes you a better-informed client.
Why is the digitalisation of the legal sector so slow? Lawyers, alongside doctors, are among the most conservative professionals. During the pandemic, Covid test results were sent to the Swiss federal authorities by fax, causing chaos. It is not legal service providers who push legal tech but their clients, who demand faster, more transparent and, above all, cheaper services. User experience is decisive. Providers that miss the legal tech wave will disappear into niches or altogether.
A personal anecdote
When we launched a purely online legal consultation in 2000 as young lawyers, the Bar Association of the Canton of Bern filed a disciplinary complaint against us with the Bar Association of the Canton of Luzern. In their view, online consultation was unprofessional and unworthy of a lawyer. Twenty years later, I advise mainly online, and I am not the only one. My clients simply want their problems solved and have no time to come and see me. The Luzern lawyers were evidently more progressive than their Bernese colleagues and dismissed the complaint out of hand.
Legal aspects of AI
There are numerous and constantly new legal issues relating to the use of AI. The most relevant for businesses, with references to the details on intla.ch and digilaw.ch (German; use the translation function):
- AI and copyright: digilaw.ch 08.05.03 Copyright and AI; digilaw.ch 08.05.04 ChatGPT & Co. and copyright
- AI and data protection: intla.ch 08 Data Protection
- Liability for AI: digilaw.ch 09.12 Liability for artificial intelligence
- AI regulation: digilaw.ch 15.02 EU Artificial Intelligence Act (AI Act); digilaw.ch 15.03 Council of Europe AI Treaty
- AI and ethics: digilaw.ch 12 Digital Ethics
AI regulation worldwide: Approaches differ widely. The EU regulates AI by risk category (AI Act, in force since August 2024, obligations phased in until 2027); its rules also affect foreign companies selling into the EU. The Council of Europe (46 member states; not the EU) has adopted the first international AI treaty, open also to non-member states; the USA, Canada, Japan and Israel are among the signatories, but the treaty is not yet in force. The USA have no federal AI law; rules come from sector regulators and individual states. China has no comprehensive AI law but a layered set of administrative rules: since 2023, providers of generative AI services to the Chinese public must register with the Cyberspace Administration of China (CAC), pass a security assessment, ensure that content complies with «socialist core values» and, since September 2025, label AI-generated content (Interim Measures for Generative AI Services, English translation; overview: White & Case, AI Watch China). Services provided from abroad to users in China must meet the same requirements or are blocked. Note for students from mainland China: Claude, ChatGPT & Co. are not available there; registered Chinese services such as DeepSeek, Qwen (Alibaba) or Ernie Bot (Baidu) are the local equivalents. Switzerland follows a sector-specific approach and signed the Council of Europe treaty in March 2025.
Check your jurisdiction
Ask a chatbot of your choice (Claude, ChatGPT & Co.): «Does [my country] have laws or binding rules on artificial intelligence that apply to companies? Which authority is in charge? Give me the sources.»
Validate the answer against the official source (statute, government or regulator website).
In class: we will compare your findings and place your countries on the map between the EU, US and Chinese approaches.
Claude, ChatGPT & Co. in the legal field
Which chatbot is best for legal questions? Just as there is no «best chatbot» in general, there is none for legal questions. It is advisable to use several chatbots in parallel and to compare their answers. You do not need a subscription for every chatbot, but one paid subscription is usually worth the price. Law firms increasingly use specialised legal AI products (see above); for business users, the general chatbots are sufficient.
Validate every output. Chatbots are not error-free: they can «hallucinate», i.e. invent plausible-sounding answers, cases or sources. Even lawyers have been sanctioned by courts for filing briefs with invented precedents, first in the USA in 2023 (Mata v. Avianca), and professional bodies now require lawyers to verify AI output (e.g. New York City Bar, Formal Opinion 2024-5). For you, a simple working rule: (1) always ask for the sources; (2) open the sources and check that they exist and say what the chatbot claims; (3) check the date; (4) if the matter is important, ask a second chatbot and, if in doubt, a lawyer. Most chatbots let you set the instruction «always cite your sources» once, in the personalisation settings.
Keep company data confidential. Never enter business secrets, customer data or personal data into a public chatbot: your input may be stored and used for training. Use the enterprise or team versions that your employer provides, which exclude training on your data, or anonymise the facts before you prompt (change names, amounts and dates). For lawyers, this is a matter of professional secrecy; for companies, of trade secrets and data protection law (Chapter 08).
Claude, ChatGPT & Co. make law borderless. Neither the country nor the language matters. You can ask which rule of Chinese law governs a contract, and the answer comes in your own language. Ask for the original source and have it translated back for verification. If the matter is important, consult a local lawyer, for instance a partner firm in your own country.
Study of legal texts, e.g. court decisions. Chatbots do not replace reading a legal text; you must still read it yourself. But they support the analysis: you can upload or paste the text and ask specific questions about it, have difficult passages explained, or check whether you have understood the reasoning correctly. Example with a decision of the Swiss Federal Supreme Court on the question whether Uber drivers are employees:
Exercise: the Uber ruling
1. Open the decision 2C_34/2021 of 30 May 2022 (Uber Switzerland GmbH and Uber B.V. v. Canton of Geneva; alternative source: juricaf.org), read it, and copy the text. The decision is in French; the chatbot will nevertheless answer in your language.
2. Prompt (Claude, ChatGPT & Co.): «I would like to discuss the following Swiss Federal Supreme Court ruling with you.» Paste the text.
3. Ask: «What are the decisive criteria under Swiss law that make an Uber driver an employee, and what does this ruling mean for a platform business model?»
4. Then ask: «How would this case be decided in [my country]? Compare the criteria.» The ruling itself compares Swiss law with decisions of the UK Supreme Court, the French Cour de cassation and the Court of Justice of the EU.
In class: in groups, each group uses a different chatbot; we compare the answers and check them against the ruling.
Drafting of contracts. Chatbots can produce checklists, sample contracts and, on the basis of concrete facts, a tailored draft. The more precisely you describe what you want and the facts of the case, the better the draft. How to draft a complete contract with a chatbot, and how to do it for one side of the table, is the exercise in Chapter 03.07.
Exercise: two checklists
Ask a chatbot of your choice (Claude, ChatGPT & Co.): «Please make me a checklist of the points a software licence agreement (Software as a Service) should regulate.» Then ask a second chatbot the same question and compare the two lists: what does one have that the other lacks, and which points would you add yourself?
In class: we collect the differences and see which chatbot forgot what.