Where does legal information end and legal assistance begin?

Law firms find themselves in a dilemma: they want to make the most of artificial intelligence, but at the same time, more and more people are entrusting their legal matters to large language models. And this, as experts point out, entails countless risks.

This issue, in the context of unprofessional legal advice, is explored in the latest industry review, «Innovations and trends in the legal sector», published by the Aranzadi LA LEY Foundation, one of the key players in the Spanish market for legal information products and legaltech. To gauge the scale, suffice it to say that the phrase «artificial intelligence» appears 125 times across the document’s 170 pages.

The institutional advocacy group is sounding the alarm over the growing number of legal advisors and individuals providing consulting services without proper education or membership in a professional organization, reports the newspaper El País. And «the emergence of artificial intelligence and generative self-learning systems has caused a significant disruption in the field of legal consulting», acknowledges Eugenio Ribón, Dean of the Madrid Bar Association (Ilustre Colegio de la Abogacía de Madrid, ICAM).

In response to this issue, ICAM submitted a proposal to parliament to impose legal restrictions on chatbots. Specifically, the idea is to amend Article 403 of the Criminal Code to establish criminal liability for providing legal consultation systems for commercial purposes without direct and effective professional supervision.

The review provides assessments of various aspects of AI use by advocates. Experts agree that caution is warranted, as most such assistants generate responses based on data from the Internet. Ensuring that this data is up-to-date, valid, and accurate is extremely difficult. One AI system may excel at generating text but be limited when it comes to ensuring legal accuracy; another may be excellent at classifying documents but unable to interpret nuances.

It is noted that large international law firms can no longer limit themselves to traditional consulting — they must design, develop, and utilize legal technologies. Technology will no longer be merely a supporting tool or communication channel; it will revolutionize legal relationships themselves, and thus — business.

However, large language models offer a level of reliability that is, at best, questionable. A solution could be training the machine on proprietary sources and documents in a secure environment. After all, the quality of AI’s responses depends on its information sources. If they are reliable, well-structured, and updated in line with changes in legislation, the model will provide safer and more accurate analytical conclusions.