The EU AI Act requires organisations that provide or use AI systems to take measures to ensure their staff have AI literacy. For law firms, legal departments and public authorities, this is primarily a question of good organisation.
Last updated: · Editorial team, German Legal Tech Hub
ContextWhat it's about
Article 4 of Regulation (EU) 2024/1689 has applied since 2 February 2025. It refers to the knowledge needed to use AI systems competently and to recognise their opportunities and risks – tailored to prior knowledge, tasks and purpose of use.
The regulation does not prescribe a specific format, such as a certificate. What matters is that the measures taken are suited to your own use and can be demonstrated.
Since July 2026, the provision has been clarified: under Regulation (EU) 2026/1744, providers and operators must take measures to promote the AI literacy of their staff; they are not required to guarantee a particular level for individual employees. The obligation itself remains in place – and since 2 August 2026, the national market surveillance authorities have been able to monitor compliance with it.
DeadlinesThe timetable of the AI Act
1 August 2024The AI Act enters into force.
2 February 2025AI literacy (Art. 4) and the prohibition of certain practices (Art. 5) apply.
2 August 2025Obligations for providers of general-purpose AI models apply.
27 July 2026Amending Regulation (EU) 2026/1744 enters into force; Art. 4 is clarified and high-risk deadlines are postponed.
2 August 2026Transparency obligations (Art. 50) apply; supervision and enforcement by national authorities begins.
2 December 2027Obligations for high-risk systems under Annex III, for example in recruitment, the judiciary and public administration.
2 August 2028Obligations for high-risk AI in products under Annex I.
In Germany, the Federal Network Agency (Bundesnetzagentur) is generally the competent market surveillance authority (§ 2 KI-MIG). Where public bodies of the federal states use AI, the authorities designated under state law are responsible.
Use casesFour steps to building AI competence in teams
1 · Stock-take: Where is AI already in use?
Starting point AI often arrives not as a standalone project but via an update: in word processing, document management, research databases or translation services. Many teams use AI without calling it that.
How technology helps A simple register records which systems with AI functions are in use, who uses them, and for which tasks.
Benefit: a foundation for training, policies and contracts – plus initial evidence that the organisation has engaged with the topic.
Requirements: queries to IT and providers, plus a brief check within the team.
Limits: The register goes out of date quickly. Decide who maintains it and when it is reviewed.
2 · Training by role rather than one-size-fits-all
Starting point A general introductory presentation reaches everyone – but no one exactly where the risks lie.
How technology helps Training is tailored to roles: lawyers learn to check results and sources; assistants and secretarial staff learn which data must not go into a tool and how to anonymise it; management learns how to select providers and conclude contracts under § 43e BRAO.
Benefit: Knowledge that's actually needed day to day, and fewer errors in data entry and review.
Requirements: a role overview and learning formats that fit the time available – live, self-study, or exchange with other organisations.
Limits: The AI Act does not require a certificate, and knowledge does not have to be measured. Training does not replace a case-by-case review.
3 · Agree a usage policy
Starting point Without clear rules, each person decides for themselves which tool to use and what to input. That’s risky when it comes to client or administrative data.
How technology helps A brief policy sets out which tools are approved, which data must not be entered, which review steps apply before a result is used, and who to contact with questions.
Benefit: Day-to-day certainty and a consistent approach towards clients, citizens and regulators.
Requirements: coordination with data protection, IT and – in companies and public authorities – the staff council.
Limits: A policy without training remains just paper. It must be updated whenever new tools are introduced.
4 · Documenting measures
Starting point When it really matters – a complaint, a regulator’s inquiry, a liability case – someone asks: what did you do about it?
How technology helps Records are kept of the register, the policy, training sessions with dates and participants, and proof of attendance for external courses.
Benefit: Traceability for regulators, clients and insurers.
Requirements: a central repository and a responsible person. According to the European Commission, internal documentation is sufficient; a dedicated AI officer or committee is not required.
Limits: Documentation does not replace genuine competence. From December 2027, anyone using high-risk systems will also need people with the necessary competence for human oversight (Art. 26(2) AI Act).
Academy
Learning what's needed for everyday work.
The Hub's Academy offers training in AI and legal tech, both live and self-paced, with a certificate of attendance under § 15 FAO. From the current programme, “Legal Tech Fundamentals” and “AI in Legal Advice” are well suited to building AI competence. The 2027 course programme is coming soon; members receive CPD hours depending on their package and 10% off further courses.
Podcast: Anisja Porschke (AI Legal Club) on training as the key to successful AI adoption.
Use case
Building AI competence within your law firm team
Starting question
A mid-sized law firm is introducing AI tools and wondering how the team can learn to use them safely – and what the AI Act requires.
Contribution from the Hub
Academy formats convey the fundamentals; in focus groups and roadshows, the team meets law firms that have already taken this step.
Possible outcome
A training plan tailored to your own tools, plus points of contact for day-to-day questions.
Anisja Porschke
“We mustn't focus solely on the cutting edge.”
Artificial intelligence is transforming the legal market at a rapid pace. While some law firms and legal departments are already working intensively with AI, many lawyers are still facing a very different question:
Where do I even begin?
Anisja Porschke, qualified lawyer and co-founder of the AI Legal Club, talks about this in the latest episode of Legal Tech Talk.
In conversation, topics include, – why training plays a key role in the successful introduction of AI, – why the Legal Tech industry needs to bring the wider legal profession along – not just talk about the latest features, – and why communities, networks and personal exchange remain more important than ever, despite all technological developments.
Because the future of law is shaped not by technology alone, but by people who are willing to learn from one another and forge new paths together.
Many thanks to Sebastian Bluhm and KSB INTAX for the professional production of our podcast series.
Interview: Sebastian Bluhm · Produced by KSB INTAX
Quick answersFrequently asked questions about AI competence
Do we need a certificate to comply with Art. 4 AI Act?
No. According to the European Commission’s questions and answers, no certificate is required, and the knowledge does not need to be measured. It is sufficient to document the measures taken internally. A certificate or proof of attendance can still be a useful building block.
Did the 2026 amendment to the AI Act abolish the obligation to ensure AI competence?
No. Regulation (EU) 2026/1744 clarified Art. 4: providers and deployers must take measures to promote the AI literacy of their staff; they do not need to guarantee a specific level for individual people. The obligation has applied since 2 February 2025.
Does Article 4 also apply if we only use a standard program with an AI feature?
Yes, if you use the AI system professionally under your own responsibility, you are a deployer within the meaning of the Regulation (Art. 3 No. 4 AI Act). The scope and depth of the measures required depend on the use case and the level of risk – a translation function requires less than a system that drafts pleadings.
Who monitors compliance in Germany?
Generally, the Federal Network Agency (Bundesnetzagentur) as the market surveillance authority (§ 2 KI-MIG). Where public bodies of the German federal states use AI, the authorities designated under state law are responsible. Oversight and enforcement by the national authorities began on 2 August 2026.
Sources and further references
That's what we discuss at the Hub.
In the Academy, focus groups and Roadshows, teams learn from organisations that are already using AI.