Malcolm Che, Head of Innovation at Identity Global
Malcolm Che, Head of Innovation at Identity Global, argues that the events and experiential marketing sector has adopted AI at speed while ignoring the governance obligations that arrived with it. With a legal deadline ten weeks away, the exposure belongs not only to agencies, but to every client who briefed them.
A synthetic, AI digital-twin version of a senior executive appears live on stage while the audience watches a keynote address. Nothing in the programme says it is AI. No disclaimer appears on the screen where the avatar holds a conversation with the human host. This happened, and it will happen again this autumn. From 2 August 2026, the regulatory framework governing that moment has repercussions.
The events and experiential marketing sector has spent two years adopting the new shiny ‘thing’ into workflows and productions. It has given far less consideration to the disclosure and compliance surrounding that, and the gap is about to get a price tag to it.
What Article 50 actually requires
The EU AI Act is a graduated compliance framework. Most of its weight falls on developers of AI systems, namely the companies building and selling the models. Article 50 is different, and applies to any organisation deploying AI in specific ways, regardless of who built the underlying tool. Most agencies, and their clients, are deployers.
Three obligations apply directly to this sector. Any AI interaction a person might mistake for human contact must be identified as AI. That includes the chatbot in your event app handling registration queries, the digital assistant in your attendee onboarding flow, the networking tool recommending connections. They are now standard features of event technology and not edge cases, with the majority running without any disclosure.
AI-generated images, video, and audio must carry a machine-readable watermark at point of creation. Deepfakes and AI-generated content on matters of public interest require explicit labelling. Between them, these obligations cover most of what agencies now produce without thinking twice: synthetic keynote talent, AI-generated visuals, voiceovers built from models rather than recorded in studios.
One distinction worth understanding: providers of generative AI tools already on the market before 2 August have until 2 December to complete watermarking implementation in their systems. That grace period belongs to the tool developers. Deployers, meaning the agencies and clients using those tools, have no such extension; the obligation begins 2 August.
Why AI in event management moved faster than the rules
Initially, there was an argument where the sector could claim ignorance, but that window is closing rapidly. Working with emerging technology has always carried an obligation beyond operational fluency. It requires thinking forward: about what the work produced, who experiences it, and what the consent relationship between content and audience actually looks like. Tech for tech’s sake is a mistake.
The problem runs deeper than production pipelines. DLA Piper, advising organisations on EU AI Act compliance, found that in many businesses there is a rush to implement AI use cases that are in some instances not even known by the compliance department. Agencies face the same internal blindspot, where AI is embedded in workflows, supplier tools, event platforms, and client-facing technology that nobody has inventoried against a disclosure framework. You can’t comply if you don’t have an inventory, enterprise licences, and technology audits.
Transparency has always been the precondition for informed consent: an audience experiencing a synthetic keynote without knowing it cannot make a considered judgement about what they are watching. Article 50 has codified a principle around ethical consideration the industry should have applied without being told; the question worth asking is why it did not.
The client carries the liability too
Article 50 obligations attach to deployers, not only producers. A brand that commissions, approves, and broadcasts undisclosed synthetic media is a deployer. The liability does not sit solely with the agency that built the content. Fines under the EU AI Act reach 35 million euros or 7% of global annual turnover, whichever is higher. That number has a way of concentrating attention in procurement.
Procurement teams at major organisations have already noticed, with AI governance questions appearing in supplier vetting processes across financial services, professional services, and the public sector. The agencies that can’t answer them clearly are losing ground at pitch stage, before a single piece of work changes hands.
The technical infrastructure is also moving faster than the sector’s awareness of it. At Google I/O on 19 May, Google confirmed that SynthID, its invisible AI watermarking system, would expand into Google Search and Chrome, with OpenAI, Kakao, and ElevenLabs adopting the standard. The system has now watermarked over 100 billion pieces of content. The argument that watermarking is technically onerous has not been credible for some time, and the tools are already doing it at point of creation. Machine-readable watermarking and human-readable disclosure is vital.
Three things, before August
For agency partners, audit every AI tool in event technology being handed to clients: registration platforms, attendee apps, engagement tools. If you haven’t already, add an Article 50 disclosure check to every pre-delivery sign-off. Brief production partners on watermarking requirements before briefs as part of the standard process, not after content is delivered.
For clients, one question cuts through: what is your agency’s synthetic media disclosure standard, and can they demonstrate it against work they have already delivered? Agencies operating to the standard as a matter of practice answer without hesitation, while those preparing to appear compliant by August take longer. The difference is the answer you’re looking for.
Responsible deployment of AI is not a differentiator, it is a baseline. The agencies worth working with on this are the ones who built disclosure into their process because they understood what the work required, approach technology ethically, and have been saying so openly. They are not the ones who read this article and updated a policy document. And clients stand to carry the hefty bill of not knowing the difference.
Article 50 is an enforceable standard, which should have been there from the beginning. The deadline marks responsible adoption.
What does the EU AI Act mean for event agencies?
AI in event management is now subject to enforceable disclosure obligations under Article 50 of the EU AI Act, from 2 August 2026. Any AI interaction an attendee might mistake for human contact must be identified as AI at the point of experience. Any synthetic media — AI-generated video, audio, or imagery — must carry a machine-readable watermark and, where it involves matters of public interest, explicit human-readable labelling. These obligations apply to deployers, meaning the agencies and clients commissioning and broadcasting AI-generated content, not only the developers of the underlying tools.
Responsible AI deployment in events starts with an inventory. Agencies need to know which tools in their production stack, event technology, and supplier network generate or present AI content, and whether disclosure is built into the delivery process. Clients commissioning AI-assisted productions should ask their agency one direct question: what is your synthetic media disclosure standard, and can you demonstrate it against work already delivered? The answer determines whether you are working with an agency that understands the obligation, or one preparing to appear as though it does.
