Legal AI

The digital Omnibus changes the AI Act timetable: what law firms should review now

Regulation (EU) 2026/1744, in force since 27 July, delays the obligations for high-risk systems and softens the AI literacy duty. We go through the dates that matter to a law firm.

Editorial photo: AI Act timetable for law firms
Image generated with AI

The European artificial intelligence regulation, Regulation (EU) 2024/1689, was born with a phased timetable. In July 2026 that timetable changed. Regulation (EU) 2026/1744, known as the digital Omnibus on AI, was signed on 8 July, published in the Official Journal of the European Union on 24 July and entered into force on 27 July 2026.

The reform does not replace the Regulation: it amends specific parts. For a law firm that acts as a user of AI tools rather than a developer, it is worth knowing what has changed.

What is postponed: the high-risk obligations

According to the recitals of Regulation (EU) 2026/1744, sections 1 to 3 of Chapter III of the AI Act, which deal with high-risk systems, will now apply:

  • On 2 December 2027 for the high-risk systems of Article 6(2) and Annex III.
  • On 2 August 2028 for the systems of Article 6(1) and Annex I.

Does a firm use high-risk systems?

Annex III lists the areas of use that turn an AI system into a high-risk one. Before assuming that a tool falls outside it, compare its real use with that annex, especially if the firm uses it in internal processes that affect people, such as managing its own staff. This review should be done against the official text of the Regulation and with the provider.

Transparency in generative AI: a transitional period

According to its recital 38, the Omnibus provides for a transitional period of four months for providers of generative AI systems that placed them on the market before 2 August 2026, in relation to their transparency obligations on generated content. If the firm uses tools of this kind, it is worth asking the provider how and when they comply.

AI literacy: the duty is softened, not removed

Article 4 of the AI Act, in its original wording, said that providers and deployers would take measures to ensure a sufficient level of AI literacy among their staff. After the Omnibus, the article says they will take measures to support the promotion of that literacy. The Commission must publish practical examples.

For a firm, the change in wording is no reason to stop training the team. Knowing which tools are used, for which tasks and within which limits is still the basis for meeting the rest of the professional duties, starting with diligence and confidentiality.

Five tasks for the last quarter

  1. Tool inventory. Note which AI systems each practice area uses, who the provider is and for which tasks.
  2. Review of supplier contracts. Check whether the data entered is used to train models, where it is processed and how long it is kept.
  3. Transparency with clients and third parties. If the firm uses chatbots or generates content with AI, check with the provider which notices and labels the Regulation requires and from when.
  4. Internal training. Document the sessions and the instructions for use, even though the literacy duty has been softened.
  5. Human verification. Set out in writing that every citation, figure or reference generated by AI is checked before it is sent to a client or filed in court.

This last point is not theoretical. Spanish courts have already sanctioned lawyers for citing non-existent judgments, as we explain in our review of the cases of fake case law generated by AI.

A European rule that coexists with the national one

The AI Act is not the only rule that affects the use of these tools in professional practice. The Organic Law on the Right of Defence includes transparency of AI criteria on legal platforms among the guarantees of the right of defence. The Code of Ethics leaves the duties of confidentiality and independence intact, whatever tool is used.

For lawyers, moreover, the most immediate risk is not an administrative penalty. It is professional and disciplinary liability for an error that the tool does not prevent.

If you want to see AI solutions for law firms live, the Legal Tech Expo 2026 takes place on 18 and 19 November at IFEMA Madrid. And if you practise on your own, in the Solo and small-firm lawyers section we publish firm management guides designed for those who do not have a compliance department behind them. We follow regulation and tools in the Legal AI section.

Fuentes

  1. Reglamento (UE) 2024/1689 de Inteligencia Artificial (EUR-Lex). Accessed 18 October 2026.
  2. Reglamento (UE) 2026/1744, Ómnibus digital sobre IA (EUR-Lex). Accessed 18 October 2026.
  3. Texto consolidado del Reglamento de IA a 27 de julio de 2026 (EUR-Lex). Accessed 18 October 2026.
  4. Reglamento (UE) 2024/1689 en la base de datos del BOE (DOUE-L-2024-81079). Accessed 18 October 2026.
  5. Reglamento (UE) 2026/1744 en la base de datos del BOE (DOUE-L-2026-81147). Accessed 18 October 2026.
  6. Reglamento (UE) 2026/1744, texto publicado en el DOUE de 24 de julio de 2026 (EUR-Lex, PDF). Accessed 18 October 2026.

The specific data in this article come from the official sources cited.

Agente Lex Newsroom Redacción

Editorial team of AGENTE LEX. We produce analysis and news for the legal profession from public and official sources, which we link at the end of each article.

View profile and articles →