欧盟《人工智能法案》重塑全球商务活动：2026年8月起实施透明度新规

Skip to content

• View Larger Image

The EU Artificial Intelligence Act officially entered into force on August 1, 2024. The regulation could significantly affect the design, sale, and delivery of meetings, congresses, exhibitions and live experiences worldwide.

Although the EU AI Act was drafted in Brussels, Belgium, its impact will reach far beyond Europe. For the business events industry, this regulation is now an operational concern for organisers, venues, destination management companies, exhibition producers, agencies, platforms, registration suppliers, audiovisual partners, and global brands.

From 2 August 2026, key AI Act requirements take effect, including transparency obligations for certain interactive and generative AI systems. The regulation applies to organisations in the EU and to companies outside the EU if their AI system’s output is used within the Union. As a result, a US-based event-tech supplier, an Asian congress organiser, or a global agency serving delegates in Europe may be subject to these rules.

The key message for the industry is clear: AI can make events more relevant, but it cannot remain invisible, unregulated, or unaccountable. Non-compliance may result in fines, loss of client trust, supplier disputes, regulatory investigations, and exclusion from risk-sensitive corporate and public-sector procurement. This is much more than a compliance issue – it fundamentally challenges the industry’s relationship with its delegates.

From useful assistant to regulated system AI has evolved from a helpful tool to a regulated system and is now integral to the event experience. It drafts invitations, translates sessions in real time, recommends networking opportunities, predicts attendance, powers registration chatbots, creates event visuals, analyses survey trends, and assists with scheduling, budgeting, and floor planning.

Many AI applications will remain permitted, as the EU AI Act is based on risk assessment rather than blanket bans. Low-risk uses have minimal obligations, while higher-risk systems face stricter requirements, and only a few practices are entirely prohibited. The Act promotes trustworthy, human-centric AI while safeguarding health, safety and fundamental rights.

This distinction matters. An AI tool that helps an event manager summarise post-event feedback is fundamentally different from systems that evaluate job applicants, use biometric data to identify delegates, infer emotions from facial expressions, or deploy virtual hosts without disclosing their artificial nature.

For meetings professionals, the key regulatory question is not, “Do we use AI?” since most organisations already do, either directly or through suppliers. Instead, it is more relevant to ask: “Where does AI make significant decisions, interact with people, generate potentially misleading content, or process sensitive personal data?”

Getting to that answer often involves a closer look at a complex supply chain. An organiser may use a registration platform from one supplier, a matchmaking app from another, AI-generated graphics from a creative agency, an AI avatar from an AV partner, and a virtual concierge from a venue. And yet, the organiser remains responsible, even if the technology is purchased rather than developed in-house.

The transparency moment Article 50 of the EU AI Act will be especially relevant for the business event sector. Its transparency requirements, effective from 2 August 2026, cover systems that interact directly with people, generative AI content, deepfakes, emotion recognition, and biometric categorisation systems.

For example, an AI concierge on an event website can answer practical questions in multiple languages, such as: “Where is the opening reception?” “Can I change my registration?” and “Which sessions are relevant to my role?” This feature significantly improves service, particularly for international congresses across time zones.

Delegates, however, should not have to guess whether they are communicating with a human or an automated system. Providers must ensure users know when they are interacting with AI. Unless this is already self-evident from the context, that disclosure must be obvious from the very first interaction and impossible to miss.

While this may seem like a very modest minimum requirement, it actually challenges the traditional industry goal of making technology seamless and “human.” Instead, event organisers must adopt a more transparent design approach, ensuring technology is both genuinely helpful and clearly identifiable.

A simple disclosure is effective: “I’m an AI event assistant and can help with registration, schedules, and venue information.” This approach sets clear expectations and lets participants decide when to involve a human, such as for refunds, safeguarding, accessibility, or more complex enquiries.

The same principle applies to virtual presenters and AI avatars. For instance, if a speaker’s appearance, speech, or voice is synthetically generated or altered, this is about far more than just checking boxes for legal compliance. In an industry built on human authenticity, deception poses profound reputational risks, and the audience’s right to stay informed must always be respected.

Synthetic content needs a provenance strategy Generative AI is transforming event marketing by quickly producing campaign images, teaser films, music, scripts, speaker introductions, social media content, and post-event highlights. For resource-constrained teams, especially independent organisers and small agencies, this speed is a significant advantage. However, the industry must differentiate between creative AI use and the misleading presentation of synthetic material.

The EU Commission’s guidance states that providers of generative AI systems must make AI-generated or manipulated audio, images, video and text detectable through effective, reliable, robust and interoperable machine-readable marking. Deployers have separate disclosure duties when people are exposed to deepfakes, and for certain AI-generated text published to inform the public about matters of public interest without human review or editorial control.

Attention to detail is essential. While not every AI-assisted image on an event website requires a prominent label, organisers should not assume they can bypass attribution simply because the material is used for marketing, where image bylines are often omitted by tradition.

A responsible event business should be able to answer four questions for each significant synthetic asset:

• Was AI used to create or materially manipulate the content?
• Could a reasonable viewer mistake it for a real person, real event, real testimonial or real occurrence?
• Has the supplier preserved the required machine-readable provenance information?
• Does the audience need a clear human-readable disclosure to avoid deception?

Consider a destination campaign featuring a crowded convention centre, engaged delegates and a renowned keynote speaker. If these scenes are artificial, the issue extends beyond Article 50. Buyers, sponsors and attendees may view the material as misleading. A brand that has built credibility over decades can lose it quickly.

This concern also applies to highlight reels produced before an event, AI-generated testimonials, fabricated press photos, cloned speaker voices, and simulated crisis communications. The key issue is not whether the generative content is technically convincing or good enough, but whether it may create a false sense of authenticity that misleads the audience.

Where event AI becomes high-stakes The industry should not assume all event-related AI tools are low risk. Certain tools may pose greater risks depending on their functions and applications.

AI-enabled networking and matchmaking illustrate these concerns. Recommending sessions based on the specific interest boxes delegates choose to check is typically low risk. However, risks increase significantly if systems infer sensitive traits, predict personal characteristics from behavioural data, rank individuals in ways that affect opportunities, or use opaque profiles to determine access to investors, VIP areas, speakers, or recruitment opportunities.

Facial recognition technologies also require caution. Even when used for basic access control, these systems may raise data protection, security, and discrimination concerns before considering the EU AI Act. If they categorise individuals using biometric data or infer emotions or personal characteristics, the legal and ethical implications grow.

One bright red line is especially relevant to employers. The AI Act prohibits systems used to infer a person’s emotions in workplaces and educational institutions, except for narrowly defined medical or safety purposes. That prohibition has been in effect since February 2025.

Event employers should exercise caution before using tools that claim to assess engagement, stress, attentiveness, happiness, or performance by analysing faces, voices, or other biometric signals. These systems are often marketed as productivity, training, or customer service solutions. In event environments, where teams are temporary, high-pressure, and highly visible, such tools may seem attractive. However, their use may be prohibited.

Recruitment presents additional concerns. Event organisers, venues, or agencies using AI to screen CVs, rank candidates, or make hiring decisions should not assume these tools are neutral. AI used in employment may be classified as high risk under the Act, with requirements that differ significantly from standard marketing automation. These systems must be documented, monitored, accurate, subject to human oversight, and free from discriminatory outcomes.

Non-compliance is a business risk Breaches of the AI Act carry severe financial penalties. Violating prohibited practices may result in administrative fines of up to €35 million or 7 per cent of total worldwide annual turnover, whichever is higher. Other violations can incur fines of up to €15 million or 3 per cent of worldwide turnover. Providing incorrect, incomplete, or misleading information to authorities may lead to fines of up to €7.5 million or 1 per cent. Although the framework considers proportionality for smaller businesses, all organisations must comply.

For global meeting companies, worldwide turnover is significant. Even if a European event is only a small part of a multinational group’s operations, a serious infringement can expose the entire group to penalties.

However, fines are only one concern. The meetings industry depends on trust. Clients, including pharmaceutical companies, government bodies, associations, and international brands, may be unwilling to justify situations where delegates unknowingly interacted with an AI agent, a speaker’s likeness was altered, or biometric profiling occurred at an event.

Potential commercial consequences include:

• Lost tenders, especially where buyers require responsible-AI assurances.
• Contract disputes among organisers, technology platforms, and creative suppliers.
• Insurance and indemnity challenges following a synthetic-content incident.
• Damage to speaker, sponsor and destination relationships.
• Reputational harm amplified through social media.

Related issues may also arise under privacy, consumer protection, employment, and intellectual property regulations.

The European AI Act does not replace the General Data Protection Regulation (GDPR). A tool may comply with one set of requirements but still present challenges under the other. For example, a transparent AI chatbot may meet AI disclosure obligations but process personal data in a way that requires a lawful basis, clear privacy information, data minimisation, and an appropriate vendor agreement.

Given this overlap, the technology team should not manage AI governance alone. Effective oversight requires input from operations, marketing, HR, procurement, legal, security, and event experience leaders.

A global standard in practice The AI Act is European law, not a global treaty, but its influence will likely extend internationally because business events operates across borders. Congresses attract global participants, brands coordinate campaigns in several markets, and event tech platforms are designed for broad deployment.

Non-European suppliers must choose between maintaining separate products and procedures for Europe or adopting European standards from the outset. Many will opt for the latter, especially when clients require consistency across international event portfolios.

The AI Act applies to organisations based in the EU and to those outside the EU if their AI system’s output is used within the Union. For example, a Singapore-based provider offering an AI networking assistant for a congress in Barcelona, or an American agency delivering synthetic keynote content to a European audience, would still be subject to these requirements.

This “Brussels effect” is particularly significant in the events sector because of procurement practices. Major corporate and association clients increasingly require suppliers to disclose data practices, security controls, and sustainability policies. Responsible AI is fast becoming a standard expectation. Agencies that demonstrate transparency, vendor governance, and human oversight may not merely avoid regulatory pitfalls, but could also secure a competitive edge in the process.

What organisers should do now  

• Start by creating an AI inventory List all AI systems used across planning, marketing, registration, attendee engagement, onsite delivery, content production, security, recruitment, and post-event analysis. Include both standalone AI products and tools with integrated AI features.
• Next, assign ownership for each system Identify the provider, deploying organisation, data involved, affected audience, country of use, contractual safeguards, and the person responsible for operation. Do not accept vague claims of “AI Act compliance”; request clear explanations of how compliance applies to each specific feature.
• Third, prioritise high-visibility attendee touchpoints Review chatbots, avatars, translation tools, facial or biometric systems, matchmaking engines, and AI-generated media. Notify attendees when they interact with AI and review synthetic content before publication.
• Fourth, ensure human intervention is available Delegates must be able to contact a person if an AI system cannot resolve an issue safely, fairly, or to their satisfaction. Human fallback is not a sign of AI failure; it is an essential part of service design.
• Finally, ensure all purchasers and users of these tools receive appropriate training The Act requires AI literacy for providers and deployers, tailored to each individual’s knowledge, experience, education, and training. Event professionals do not need to become machine learning engineers, but they should understand each tool’s capabilities, limitations, potential risks, and when to escalate concerns.

Future events will be more intentional, not less technological. Success will belong to organisers who combine creativity with transparency, efficiency with human judgement, and innovation with respect for the people whose attention, data, and trust make events possible.

Page load link

Go to Top