Spanish legaltech changes its pitch: no longer who writes fastest, but who lets you check what it says
Spain's legaltech sector reorganises around rigor and verifiable sources, with confidentiality and GDPR compliance as central demands.
Generative AI is already part of the daily routine of many Spanish law firms: it summarises files, prepares first drafts, orders chronologies and finds case law in seconds. But the question running through the profession is no longer about speed — it is about trust: can a lawyer rely on what a machine tells them?
Adoption outruns training
Recent data paint a profession that has jumped without a net. The third edition of the Radiography of Digital Transformation in Law Firms, a joint work by El Confidencial and consultancy LOIS, detects real progress across seven areas, from internet visibility to technological resources. And the White Paper on Artificial Intelligence and the Legal Profession, promoted by the General Council of Spanish Lawyers, adds the contrast: almost two in three surveyed lawyers admitted having no training in these tools. Adoption moves faster than learning, and that mismatch is where the problems begin.
The sector's answer: sources or nothing
Faced with this panorama, the Spanish legaltech sector has settled on one idea: a useful legal AI is not the one that writes fastest, but the one that lets you verify what it says. The debate has moved from productivity to rigor and sources, as an analysis published by El Independiente notes.
Its example is Maite AI, an assistant specialised in Spanish law whose proposal starts from a simple principle: every answer comes with the original source so the lawyer can review it before using it. The tool targets the firm's everyday tasks: analysing files and court records, building chronologies of facts, preparing the grounds for claims and appeals, spotting weaknesses in the opposing party's submissions or stress-testing legal arguments before filing.
The approach changes the relationship between professional and machine: AI acts as a first filter and leaves verification in the lawyer's hands. That is precisely what was missing in the sanctioned cases, the analysis recalls: lawyers trusted references that looked correct and never checked them in CENDOJ, the free public case-law database.
The other demand: confidentiality
Uploading a client's file to a generalist assistant raises serious doubts about professional secrecy. Solutions aimed at the sector try to respond: in the case of tools for law firms like the one cited, the public commitment is not to use client data to train models, to host servers in Europe and to comply with the GDPR and Spanish data-protection law. For many firms, this guarantee already weighs as much as the quality of the answers.
2027, the year of order
The regulatory calendar is tightening. The transparency obligations of the European AI Regulation apply from August 2026, and the Digital Omnibus, approved in July, postponed to 2 December 2027 the requirements for Annex III high-risk systems, which include those used in the administration of justice. Providers and professional users have little more than a year to prepare documentation, human oversight and controls.
The ethical framework is also taking shape: the General Council of the Judiciary has approved instructions on judges' use of AI, and in the first Study on Digital Justice by the Madrid Bar Association, 75.7% of respondents supported some form of obligation to declare AI use in proceedings. The conclusion: 2027 will be the year the debate moves from whether to use AI to using it well, and the advantage will belong to whoever proves they control it.
Internal links
Fuentes
- elindependiente.com. Accessed 29 October 2026.
The specific data in this article come from the official sources cited.
