Skip to content
Blentera

Resources, EU AI Act

The EU AI Act for companies using AI agents

Most companies that use AI agents are deployers under the EU AI Act. This guide sets out what that role requires today, what changes with the Digital Omnibus, and what to do first.

Updated 20 November 2026. 9 minute read.

In short

  • A company that uses an AI agent under its own authority is a deployer. It becomes a provider if it puts its name on a high-risk system, makes a substantial modification, or turns a general-purpose agent to a high-risk purpose (Article 25).
  • Prohibited practices (Article 5) and the AI literacy duty (Article 4) already apply. Transparency duties (Article 50) apply from 2 August 2026.
  • The Digital Omnibus on AI, Regulation (EU) 2026/1744, entered into force on 27 July 2026. It moved the high-risk obligations for Annex III use cases, such as employment, to 2 December 2027, and for Annex I products to 2 August 2028.
  • For high-risk use, deployers must follow the provider's instructions, assign trained human oversight, monitor the system, keep logs for at least six months and inform workers before use at the workplace (Article 26).
  • Fines reach up to €35 million or 7% of worldwide annual turnover for prohibited practices, and up to €15 million or 3% for most other breaches.

Provider or deployer: which role you hold

The AI Act does not use the word agent. It regulates AI systems, and an AI agent that plans and takes actions is an AI system in the sense of Article 3(1). What you must do depends on your role.

A provider develops an AI system, or has one developed, and places it on the market or puts it into service under its own name. A deployer uses an AI system under its authority in a professional activity (Article 3(4)). A company that buys or builds agents and lets them draft emails, triage tickets or update records is, in most cases, a deployer. The model vendors and agent platforms are usually providers.

When a deployer becomes a provider

Article 25(1) treats a deployer as the provider of a high-risk AI system, with all provider duties, in three cases:

  • it puts its name or trademark on a high-risk AI system already on the market;
  • it makes a substantial modification to a high-risk AI system so that it remains high-risk;
  • it changes the intended purpose of an AI system that was not high-risk, including a general-purpose AI system, so that it becomes high-risk.

The third case matters most for agents. A general-purpose agent pointed at a high-risk task, such as filtering job applications, can make your company the provider of a high-risk system. Provider duties include a risk management system, technical documentation and a conformity assessment. Check this before you give an agent a new job.

Keep a short written record of each agent use case, its purpose and your role. It is the first thing a regulator, auditor or customer will ask for.

What applies when, as of November 2026

The AI Act is Regulation (EU) 2024/1689. It entered into force on 1 August 2024 and applies in stages. The Digital Omnibus on AI, Regulation (EU) 2026/1744 of 8 July 2026, amended the Act and entered into force on 27 July 2026. The Council gave its final approval on 29 June 2026.

  • 2 February 2025: prohibited practices (Article 5) and AI literacy (Article 4) apply.
  • 2 August 2025: rules for general-purpose AI models, governance and penalties apply.
  • 2 August 2026: most remaining rules apply, including the transparency duties in Article 50.
  • 2 December 2026: the new prohibition on systems that generate non-consensual intimate imagery or child sexual abuse material applies, and the grace period ends for marking AI-generated content from generative systems already on the market before 2 August 2026.
  • 2 December 2027: high-risk obligations apply to the stand-alone use cases in Annex III, including employment, education, credit scoring and access to essential services.
  • 2 August 2028: high-risk obligations apply to AI in products covered by the Union safety laws in Annex I, such as machinery, lifts and toys.

Before the Omnibus, the Annex III obligations were due on 2 August 2026. The delay gives time for harmonised standards and guidance. It does not remove the obligations or change their content, so the preparation work stays the same.

Member States may still adopt their own rules on points the Act leaves to them, such as penalties. Check national guidance where you operate.

Prohibited practices and AI literacy

Article 5 bans a short list of AI practices outright. For a company using agents, the relevant bans are:

  • manipulative or deceptive techniques that materially distort behaviour and cause significant harm;
  • exploiting vulnerabilities due to age, disability or a social or economic situation;
  • social scoring that leads to unjustified or disproportionate detrimental treatment;
  • inferring emotions of individuals in the workplace or in education, except for medical or safety reasons;
  • biometric categorisation to infer sensitive characteristics such as political opinions or sexual orientation;
  • untargeted scraping of facial images to build facial recognition databases;
  • from 2 December 2026, systems that generate non-consensual intimate imagery or child sexual abuse material.

An agent that reads employee chat messages to score mood or stress at work would fall within the emotion recognition ban. The Commission's guidelines on prohibited practices give worked examples for each ban.

AI literacy (Article 4)

Providers and deployers must take measures to support the AI literacy of their staff and others who operate or use AI systems on their behalf. The Omnibus rewrote this duty. It no longer requires a guaranteed level of literacy for each individual, but it still applies today and it covers every company that uses AI agents, high-risk or not.

In practice, this means role-based training: what the agents do, their limits, how to read their output and when to stop them. Keep a record of who was trained and when.

Transparency duties under Article 50

Article 50 applies from 2 August 2026 and is not affected by the high-risk delay. Some duties sit with providers and some with deployers.

  • Providers must design AI systems that interact directly with individuals so that those individuals know they are dealing with an AI system, unless it is obvious.
  • Providers of systems that generate synthetic audio, images, video or text must mark the output in a machine-readable way.
  • Deployers of emotion recognition or biometric categorisation systems must inform the individuals exposed to them.
  • Deployers that publish deep fakes must disclose that the content is artificially generated or manipulated.
  • Deployers that publish AI-generated text to inform the public on matters of public interest must disclose it, unless the text has gone through human review or editorial control and an employee or the company holds editorial responsibility.

For agents, the common case is a customer-facing agent that answers chats or emails. If you build that agent yourself, you may carry the provider duty to make clear it is an AI. Even where you do not, telling customers plainly is the safe default.

When an agent use case is high-risk

A use case is high-risk if it falls within one of the areas listed in Annex III, unless an exception applies. The areas include biometrics, critical infrastructure, education, employment, access to essential private and public services, law enforcement, migration and the administration of justice.

Employment, the area most companies meet first

Annex III point 4 covers AI systems used to recruit or select candidates, including placing targeted job adverts, filtering applications and evaluating candidates. It also covers systems used to make decisions on promotion, termination or terms of work, to allocate tasks based on individual behaviour or traits, and to monitor or evaluate performance and behaviour.

An agent that ranks CVs, scores interviews or flags employees for performance reviews is therefore likely high-risk. An agent that books interview rooms is not.

The Article 6(3) exception

An Annex III system is not high-risk if it does not pose a significant risk of harm. This covers narrow procedural tasks, improving the result of a completed human activity, detecting patterns without replacing human assessment, and preparatory tasks. The exception never applies where the system profiles individuals. The provider must document this assessment.

Deployer duties for high-risk systems

From 2 December 2027, a company that deploys a high-risk Annex III system must meet Article 26. The main duties are:

  • use the system in line with the provider's instructions for use, with appropriate technical and organisational measures;
  • assign human oversight to employees with the necessary competence, training and authority, and give them the support they need;
  • ensure input data under your control is relevant and sufficiently representative for the intended purpose;
  • monitor operation, and where the system may present a risk, inform the provider and the market surveillance authority without undue delay and suspend use;
  • report serious incidents, first to the provider and then to the authorities;
  • keep the logs the system generates, to the extent they are under your control, for at least six months unless other Union or national law, in particular data protection law, says otherwise;
  • as an employer, inform workers' representatives and the affected workers before putting the system into service or using it at the workplace;
  • inform individuals when a high-risk system makes or assists decisions about them;
  • use the provider's information to carry out a data protection impact assessment under GDPR Article 35 where one is required.

Some deployers must also carry out a fundamental rights impact assessment before first use (Article 27). This applies to public bodies, private entities providing public services, and deployers that assess creditworthiness or price life and health insurance. Individuals affected by a decision based on a high-risk Annex III system have a right to a clear explanation of the role the system played (Article 86).

Fines

Under Article 99, breaching the prohibited practices can cost up to €35 million or 7% of total worldwide annual turnover, whichever is higher. Breaching most other obligations, including those of deployers and the transparency duties, can cost up to €15 million or 3%. Supplying incorrect information to authorities can cost up to €7.5 million or 1%. For SMEs and start-ups, the lower of the two amounts applies.

A step-by-step checklist

  1. List every AI agent in use, with its owner, purpose, data and the apps and tools it can reach.
  2. For each agent, record your role: deployer or provider. Check the three Article 25 triggers.
  3. Screen each use case against Article 5. Stop anything that comes close to a ban.
  4. Classify each use case against Annex III, and record why it is or is not high-risk.
  5. Put AI literacy training in place for every employee who works with agents, and keep a record.
  6. Make customer-facing agents identify themselves as AI, and label published AI content where Article 50 requires it.
  7. For high-risk use cases, name the employees who oversee each agent and give them the authority to stop it.
  8. Turn on logging and set retention to at least six months, in line with your GDPR retention rules.
  9. Inform workers' representatives and affected workers before any high-risk use at the workplace.
  10. Run a data protection impact assessment where personal data is involved and the risk is high.
  11. Set a review date for each use case, and review it when the agent, model or purpose changes.

Start now, even though Annex III duties apply from 2 December 2027. The inventory and classification steps take the longest and they also support your GDPR records.

How Blentera helps

Blentera is a platform where companies run AI agents with employees in charge. Each agent has a role in the org chart. Approval rules can be set per agent, project, app or tool, and budgets warn at 80% and pause the agent at the limit. Agents share one company memory, and every action lands in an activity log that can be exported. An EU AI Act use-case register checks requirements such as human oversight and review dates before an agent goes live. It is hosted in the EU. These tools support your work on the steps above. Meeting the AI Act remains your company's responsibility.

Questions

Sources

  1. Regulation (EU) 2024/1689 (AI Act), consolidated text of 27 July 2026, EUR-Lex
  2. Regulation (EU) 2026/1744 (Digital Omnibus on AI), EUR-Lex
  3. AI Act: regulatory framework and timeline, European Commission
  4. Navigating the AI Act: frequently asked questions, European Commission
  5. Council gives final green light to simplify and streamline AI rules, Council of the EU, 29 June 2026
  6. EU agrees to simplify AI rules and ban nudification apps, European Commission
  7. Transparency obligations under Article 50 of the AI Act, European Commission
  8. Guidelines on prohibited AI practices, European Commission
  9. Article 26: Obligations of deployers of high-risk AI systems, AI Act Service Desk
  10. Article 99: Penalties, AI Act Service Desk

This guide is general information, not legal advice. Check how the rules apply to your company with your own advisers.

Get started

Put your first agent to work in minutes.

Three steps, and your first agent is working for your company.

  1. Step 1

    Create an account

    Sign up with your work email and invite your team. No sales call needed.

  2. Step 2

    Add your first agent

    Create an agent, give it a role, and connect your own AI key and the apps it may use.

  3. Step 3

    Give it a task

    Ask in Blentera or your chat app, and follow the work until it is done.