Everyday Law

The scope of legal data in the scrapable sphere is broad, but not all-encompassing. “Everyday” law is missing.

Currently, legal AI is largely trained on the “tip” of the iceberg: formal laws, published case law, and academic bar exams. These documents are easy to retrieve and scrape, not copyright-protected, and are what is typically considered the “core of law”. But this is not all law is.

In this workshop, we look into datasets for legal AI and ask ourselves whether these are representative enough or not. Most conflicts, even if legal in nature or subject to legal rules, are resolved—or abandoned—long before they reach a court. Even for those that reach courts, a large majority ends up settling before resulting in a written decision. Often, issues or difficulties that could be solved via litigation or complaint are simply agreed upon amicably between parties in informal settings, over a handshake and compromise. In short, legal datasets are missing a lot of valuable information at this stage and do not represent the lived realities of everyday legal challenges. In this workshop, we will analyze how this data can be accounted for, and what value it can bring to the development of legal AI.

By only focusing on law and case law, existing legal AI systems and benchmarks thus only have access to a small section of how law interacts with citizens and society. This focus on unusual and high-stakes situations could introduce biases into AI systems.

For example, AI systems—and then their users—may get the impression that going to court is their only option in certain situations, while informal steps such as negotiation, mediation, or early-stage conflict resolution are in reality employed much more frequently. This bias could thus lead to a more litigious society that prioritizes adversity over conflict containment and resolution.

In this workshop, we will explore the question of creating more representative legal datasets, looking beyond cases and statutes. Such data, which we refer to as Everyday Law, would allow us to get a better understanding of how most people interact with the legal system and allow us to build AI approaches that better reflect these interactions.

A workshop of a similar nature took place in summer 2026 at the University of Lausanne (UNIL), where this conversation was first had. Mapping out the concept of “everyday law” was a key component of the gathering. Discussion centered on where “missing law” data could instead come from (legal aid records, advice platforms, forums, search logs, phone surveys, tacit knowledge from practitioners) and the accessibility, ethical, and institutional barriers to getting it (agencies not keeping stats, informal rejection practices, PII/privacy concerns, platforms restricting data access). A recurring theme was whether LLMs “over-diagnose” everyday situations as needing formal legal action (an analogy to inexperienced doctors over-diagnosing rare diseases), and whether fine-tuning or prompting could correct this bias without causing under-diagnosis instead. The group also debated deeper questions—whether reasoning or data is the real bottleneck for legal LLMs, whether more legal information actually empowers people or just formalizes disputes unnecessarily, and how to reach and study underserved populations.

This workshop will build on the results and discussions of the previous workshop.

We will bring together researchers who approach AI and law from different angles. On the technical side, we welcome computer scientists and NLP researchers working on language models and benchmarks. The workshop also welcomes legal scholars focused on access to justice. Because everyday law involves human behavior, we welcome sociologists, anthropologists, and psychologists who study how people recognize legal problems or settle disputes informally. Human-computer interaction researchers working on user trust and plain language are also welcome. We welcome frontline practitioners, like legal aid providers, ombudspersons, and tech developers, who actually help people navigate routine conflicts. Additionally, those whose path is adjacent to social work can be a valuable addition: tools for access to justice and law are even more useful when the institutions and support services that employ them can be equally well invested in their development.