Tomorro and Pappers: Legal AI Tested Against Sovereignty

A Strategic Partnership for Legal AI

On March 3, French legaltech companies Tomorro and Pappers announced a partnership that marks a significant step in the evolution of artificial intelligence applied to law. This partnership aims to integrate Pappers' structured legal databases into Tomorro's AI platform to enhance the reliability of responses generated by AI models. Pappers emphasizes its sovereign positioning by using data from European sources, French capital, and hosting in France.

In a sector where handling sensitive information is common, these characteristics could influence the decisions of law firms and legal departments.

This collaboration between Tomorro and Pappers is part of a broader trend in the industry. After a rapid adoption of generative AI tools, law firms and legal departments now find themselves at a selection juncture. The question is no longer limited to the performance of the models but extends to the ecosystem in which they operate: the quality of the databases, the security of the processed information, and potentially, the technological sovereignty of the solutions.

Rapid Adoption of AI in the Legal Sector

Generative artificial intelligence has quickly established itself in legal professions, facilitating text analysis and document research. According to the Future Ready Lawyer study by Wolters Kluwer, nearly three-quarters of legal professionals plan to integrate generative AI into their work over the next year, and about one-third of firms are already using it. The study reveals that 76% of firms believe these technologies will improve legal work efficiency, while 68% anticipate a positive impact on revenue growth.

However, the use of general-purpose tools like ChatGPT has revealed limitations, including hallucination phenomena, where non-existent case law is produced. Gabriel Moneyron, co-founder of Pappers Justice, emphasizes that legal security and the responsibility of lawyers are at stake. "Many professionals still use ChatGPT for legal research. The problem is that these models can produce case law that does not exist," he observes.

These hallucination phenomena are now well-documented. Researcher Damien Charlottin lists several court decisions on his website where lawyers have cited non-existent case law generated by AI tools. In some cases, judges explicitly mentioned in their rulings that the cited references appeared to come from an AI tool. For the lawyers involved, the consequences can be significant, including disciplinary sanctions. "The most critical question is not just about data confidentiality. It's also about legal security. If a legal professional relies on erroneous information produced by AI, the responsibility remains theirs," emphasizes Gabriel Moneyron.

Today, most legal services and law firms are therefore equipping themselves with verticalized AI solutions specialized in law.

Data Quality: A Major Challenge

In the field of legal AI, performance depends on data quality. The partnership between Tomorro and Pappers illustrates this trend by connecting AI to structured legal databases to improve response reliability. Large organizations are also developing their own AI tools to leverage their internal data but require reliable external sources.

Gabriel Moneyron explains that data structuring is a crucial competitive advantage, supported by significant investments in R&D. "We have been investing millions of euros in R&D for several years to structure these legal databases. This is what allows AI tools to produce much more reliable responses," he explains. This analysis is shared by Xavier Niffle, a partner at KPMG in charge of AI topics. "The quality of the response directly depends on the database from which the model retrieves the information. AI should be seen as an accelerator that allows for faster and deeper exploitation of a knowledge base," he details.

For him, these tools remain assistants and not substitutes for legal reasoning. "Generative AI allows for much quicker access to information, but it does not replace legal analysis. Human validation will remain essential in these professions," he asserts.

Technological Sovereignty in Question

With the increasing integration of AI into legal processes, technological sovereignty is becoming a concern. Some European legaltech companies prioritize local infrastructures and European data. Gabriel Moneyron notes a shift in client expectations, with a growing demand for European solutions.

However, according to Xavier Niffle, sovereignty is not yet a decisive criterion for companies, which prioritize efficiency. "Companies are primarily looking for the most effective solutions. Sovereignty is becoming a topic, but it is not yet always decisive in decision-making," he observes. Public procurement could influence this trend by favoring European solutions. "If administrations start to prioritize European solutions in their calls for tenders, it could help structure a local ecosystem and accelerate the emergence of specialized players," he estimates.

Large international firms often adopt a pragmatic approach. For example, Ashurst has deployed the American AI Harvey globally while ensuring client data is isolated. "We use Harvey in a closed and dedicated environment. Our clients' data remains within the firm's infrastructure and is not used to train the models," explains Nicolas Quoy, partner in Digital Economy at Ashurst. For him, client expectations today remain focused on the security and reliability of tools. "The priority remains data protection and the quality of results produced by the tools. The question of sovereignty exists, but it is not yet articulated by clients," he adds.

As artificial intelligence becomes central to legal production, the balance between technological performance, data reliability, and sovereignty could, however, become one of the structuring decisions in the sector.