EU AI Act timeline: which rules apply to robots and when, after the AI Omnibus

Under the EU AI Act as amended by the AI Omnibus, which entered into force on 27 July 2026, the high-risk rules now apply from 2 December 2027 for systems in sensitive areas and from 2 August 2028 for AI built into regulated products such as machinery. That second date is the one most relevant to robots. Bans and transparency duties already apply.

This article sets out the timeline, explains why robots usually fall under the product-related part of the Act, and outlines what hirers and suppliers can reasonably expect. It is general information, not legal advice. Whether a particular robot is caught depends on its design, its intended purpose and the role each company plays.

The AI Act in brief

The AI Act, Regulation (EU) 2024/1689, sorts AI systems by risk. According to the European Commission's AI Act page, there are four levels: unacceptable risk (banned practices), high risk (strict requirements), transparency risk (disclosure duties) and minimal risk, which covers most AI in use today and has no specific rules.

The Act also separates roles. A provider develops an AI system, or has it developed, and places it on the market. A deployer is any person or organisation using an AI system under its authority, apart from purely personal, non-professional use. In the events world, the robot manufacturer is normally the provider. A rental company, agency or brand running the robot at an event will often be a deployer.

The timeline as it stands in October 2026

The Commission lists the following dates on its AI Act page:

1. 1 August 2024 – the AI Act entered into force.
2. 2 February 2025 – the prohibited practices started to apply.
3. 2 August 2025 – governance rules and obligations for general-purpose AI models started to apply.
4. 2 August 2026 – the general application date, including the transparency rules.
5. 2 December 2027 – high-risk rules for systems used in sensitive areas, such as employment, education, credit scoring and biometrics.
6. 2 August 2028 – high-risk rules for AI embedded in products covered by EU product-safety law, such as machinery, medical devices, toys and lifts.

The last two dates are the result of the AI Omnibus. As originally adopted, Article 113 of Regulation (EU) 2024/1689 set two separate start dates for the high-risk rules:

• 2 August 2026 for high-risk systems in the sensitive areas listed in Annex III.
• 2 August 2027 for AI in products covered by the EU product-safety laws listed in Annex I, which include machinery.

The Commission's AI Act Service Desk FAQ describes the new dates as at most 16 months later than originally planned for the sensitive areas and at most 12 months later for products. The Commission's AI Act FAQ says the late arrival of technical standards had put the August 2026 start of the high-risk rules at risk.

What the AI Omnibus changed

The Commission published its Digital Omnibus on AI proposal on 19 November 2025 as part of a wider simplification package. Its stated aim was a smooth and proportionate rollout of some AI Act provisions. The Commission reports that the co-legislators reached political agreement on 7 May 2026 and that the amending act entered into force on 27 July 2026.

The Omnibus did not delay everything. The Commission says it left the August 2026 transparency deadline unchanged. The main change concerns the high-risk rules. According to the Commission FAQ, they are now tied to the availability of harmonised standards, common specifications or Commission guidelines, so that companies have the tools they need before the obligations bite. The Commission presents the later dates as simplification, not a weakening of the requirements themselves.

For robots this is important. The product-related rules, which cover AI in machinery, were originally due on 2 August 2027 and now apply from 2 August 2028, so the shift for robots is one year. The sensitive-area rules moved further, from 2 August 2026 to 2 December 2027.

Why robots usually sit in the product-safety part of the Act

The AI Act has two routes to high-risk status. One covers use cases listed in the Act, such as recruitment or access to education. The other covers AI that is a safety component of a product regulated by existing EU product law, or that is itself such a product, where that law requires a third-party conformity assessment. Machinery law is one of the listed pieces of legislation.

The Commission FAQ gives "AI systems operating robots" as an example of this product-based category. A humanoid or a robot dog that uses machine learning to balance, navigate or avoid people may therefore come under the AI Act as part of a machinery product, not as a standalone software tool.

This does not mean every event robot is automatically high-risk. Classification depends on whether the AI performs a safety function and whether the product law requires third-party assessment. A robot driven entirely by a human operator with a remote control is a different case from one that moves autonomously through a crowd. Suppliers and manufacturers are best placed to explain how their robot is classified. Where that is unclear, it should be put in writing.

The Machinery Regulation runs in parallel

A second deadline arrives sooner. The new Machinery Regulation (EU) 2023/1230 becomes mandatory on 20 January 2027, according to the Commission's machinery page. Until then, machinery must comply with the older Machinery Directive 2006/42/EC, although manufacturers may already declare conformity with the new regulation.

The Commission says the new regulation includes provisions for machinery whose safety functions are AI-powered, adds cyber-safety requirements for safety-relevant software and control systems, and covers the conformity assessment of higher-risk machinery. In practice, many robot makers will face the new machinery rules about 18 months before the AI Act's product-related high-risk obligations begin on 2 August 2028. Under the original AI Act timetable, that gap would have been just over six months.

Standards are still being written

The delay is linked to standards, so their progress is worth following. The Commission has asked the European standards bodies CEN and CENELEC, working through their joint committee JTC 21, to develop standards in ten areas. These include risk management, data quality, record keeping, transparency, human oversight, accuracy, robustness, cybersecurity and quality management, according to the Commission's AI Act standardisation page.

The same page reports that the quality-management standard prEN 18286 was the first to go to public enquiry, on 30 October 2025. It was later referenced in the Official Journal through Commission Implementing Decision (EU) 2026/2274. Using harmonised standards is voluntary, but products that follow them are presumed to meet the corresponding legal requirements.

What applies to robots at events today

Even before the high-risk dates, some AI Act rules already matter for robots at events.

• Transparency. Since August 2026, people who interact with a chatbot must be told they are dealing with a machine, according to the Commission's AI Act page. A robot that holds conversations through a language model is an obvious case where visitors should know they are talking to an AI.
• Prohibited practices. The bans that have applied since February 2025 include certain uses of biometric identification and social scoring. Any camera-based feature on a robot should be checked against them. Data protection law applies alongside the Act.
• General-purpose AI. If a robot's speech runs on a third-party general-purpose model, the model provider has had its own obligations since August 2025. The company integrating the model still needs to understand what it is using.

When the high-risk rules arrive, deployers will have duties of their own. The Commission lists them as using the system in line with the provider's instructions, monitoring how it performs, acting on identified risks, assigning trained people to human oversight and, in certain cases, informing people affected by its decisions. These duties are similar to what responsible robot operators already do at events.

Points that remain uncertain

Some questions cannot be settled yet. Not all harmonised standards have been published. Commission guidelines on high-risk classification and on the AI Act's relationship with product-safety law will shape how individual robots are treated. National market surveillance authorities are still building up their enforcement practice. A company that modifies a robot's software or changes its intended purpose could also take on more obligations, and that point should be discussed with the supplier and, where needed, a lawyer.

What this means for event organisers

• For robots covered as machinery products, the high-risk obligations do not start before 2 August 2028, but the transparency rules already apply. Tell visitors when they are talking to an AI.
• Ask suppliers which regime they follow. That means whether the robot is CE-marked under the Machinery Directive or the new Machinery Regulation, and whether the manufacturer has assessed it under the AI Act. Include this in your brief to suppliers.
• Keep human oversight visible. A trained operator and a clear stop procedure are good practice now and match future deployer duties. See our guide to robot operators and staffing and the safety planning guide for public events.
• Treat cameras and microphones as a compliance topic. Review the guide to cameras, microphones and visitor privacy before switching on recognition or recording features.
• Record roles in the contract. Set out who is the provider, who operates the robot and who may change its software, alongside the points in the insurance and liability guide.

Frequently asked questions

When do the AI Act high-risk rules apply to robots?

After the AI Omnibus, the European Commission states that high-risk rules for AI embedded in products covered by EU product-safety law, such as machinery, apply from 2 August 2028. High-risk rules for systems used in sensitive areas, such as employment or education, apply from 2 December 2027. Robots generally fall into the product-related group when their AI performs a safety function.

What was the Digital Omnibus on AI?

It is a targeted amendment to the AI Act that the European Commission proposed on 19 November 2025. According to the Commission, political agreement was reached on 7 May 2026 and the act entered into force on 27 July 2026. It links the start of the high-risk rules to the availability of standards and guidance, which moved those deadlines to December 2027 and August 2028.

Is every humanoid robot or robot dog a high-risk AI system?

No. High-risk status under the product route depends on whether the AI acts as a safety component of a product, such as machinery, that requires third-party conformity assessment under EU law. A remotely operated robot and one that moves autonomously among visitors may be treated differently. The supplier or manufacturer should explain the classification. This is general information, not legal advice.

Does anything in the AI Act already apply to robots at events?

Yes. Since August 2026, transparency rules require that people are told when they are interacting with an AI system such as a chatbot, which covers conversational robots. The bans on certain practices, including some biometric uses, have applied since February 2025. Data protection law applies alongside these rules whenever a robot records or analyses visitors.

How does the new Machinery Regulation relate to the AI Act?

The Machinery Regulation (EU) 2023/1230 becomes mandatory on 20 January 2027 and replaces the Machinery Directive. The Commission says it includes provisions for machinery with AI-powered safety functions and for cyber-safety. For many robots it will therefore apply well before the AI Act's product-related high-risk obligations begin in August 2028.

How this article was made: written by the humanoid4rent editorial team with AI assistance; every factual statement was checked against the sources listed below on 2026-10-31. Spotted an error? Write to info@humanoid4rent.com.