With the EU AI Act's transparency obligations under Article 50 now in the enforcement spotlight, organizations are bracing for a new era of regulatory oversight. The first year of enforcement is expected to shape how companies deploy and govern AI systems, particularly those that interact with natural persons or generate synthetic content. In a recent interview, Edwin Weijdema, Field CTO at Veeam, shared his perspective on what that first year may bring, drawing on parallels from GDPR and NIS2 implementation.
Article 50 exposure and enforcement realities
Article 50 breaches carry exposure up to €15 million or three percent of worldwide turnover. However, Weijdema cautions against expecting immediate large-scale financial penalties. As seen with NIS2 and GDPR, enforcement is delegated to individual member states, each with its own procedures and priorities. This makes precise prediction difficult.
Weijdema suggests that the first year of enforcement may be treated as a “bedding in” period. Corrective orders are likely to significantly outnumber major financial penalties, especially for organizations demonstrating genuine compliance efforts. Regulators will consider proportionality, scale of impact, intent or negligence, cooperation speed, and the existence of basic governance controls when deciding on actions.
While headline-making fines often appear to show regulators mean business, Weijdema does not expect such fines to land in year one. The bigger practical exposure for organizations may be operational rather than financial. Being ordered to suspend, relabel, change, or withdraw an AI-enabled process at speed can be far more disruptive than paying a fine. “In year one, the bigger risk likely won’t be the fine; it’ll be being told to stop using the system until you can prove it is compliant,” he notes.
Agentic systems and direct interaction
A key area of uncertainty involves agentic AI systems that interact with people indirectly through ticketing queues, shared inboxes, or supplier procurement portals. Weijdema clarifies that under the EU AI Act, the channel is not decisive. A ticketing queue does not automatically mean direct interaction, but it can. The critical question is whether the AI system itself is communicating with a natural person, or whether a human intermediary exercises meaningful review and control.
If an AI drafts a response and a human reviews and sends it, the risk profile differs significantly from an AI agent autonomously replying to a customer, supplier, or employee. The latter can start to look like direct interaction, even if it happens through a ticketing system or procurement portal rather than a chatbot window.
Weijdema advises companies to make deliberate choices to separate internal and customer-facing agents by setting up appropriate barriers based on their roles. Access and privacy controls should be present across the entire organization, not just across agents. “It’s all well and good telling an agent ‘don’t go into this room’; you also need to put a lock on the door,” he says.
Ultimately, the AI Act does not care whether interaction happens in a chatbot window or a ticket queue. It cares whether the human is effectively dealing with the machine.
Phishing simulations and cloned voices
Security teams often run simulated phishing and vishing exercises, sometimes cloning an executive’s voice. These exercises may fail if the material carries a transparency label. Weijdema warns that such exercises are not automatically exempt from the AI Act’s transparency requirements. Cloning an executive’s voice is especially sensitive because it can quickly create a deepfake scenario where a real person appears to say something they did not say.
A security purpose does not automatically create an exemption, and the argument that “the exercise works better without disclosure” is not, by itself, a compliance justification. Organizations deciding not to label AI-generated elements must demonstrate that the legal basis and risk of that decision have been carefully assessed.
Weijdema recommends involving legal and compliance departments early to document reasoning. Privacy, HR, and employee representative input should also be included, especially when using a real person’s voice, image, or likeness. He suggests considering alternatives such as fictional personas, synthetic voices that do not imitate real employees, prior general notice that simulations may use synthetic media, and immediate post-exercise disclosure. The goal is to preserve realism without normalizing undisclosed executive impersonation.
Documentation should cover the purpose of the exercise, scope, AI tools used, whether any real person was imitated, disclosure provided and when, personal data processed, necessity and proportionality, safeguards in place, and employee debriefing. Weijdema’s advice to security teams: “A security objective does not magically turn an undisclosed deepfake into a compliant one. If you have to clone the CEO’s voice to make the test work, legal should be in the room before anyone presses send.”
Where the first Article 50 action will originate
As of mid-June, only nine of the twenty-seven member states had designated both a market surveillance authority and a notifying authority, twelve had partial designations, and six had neither. This uneven readiness raises questions about where the first Article 50 action will come from.
Weijdema says the first action is formally most likely to come from a market surveillance authority, as enforcement responsibility sits at national level. However, the trigger may come from elsewhere. Defamation claims are possible, especially where synthetic audio or video damages someone’s reputation, but that is more likely to be a parallel legal route than the first clean Article 50 enforcement case.
Consumer groups could be likely candidates for early challenges, especially for AI systems that affect or interact with large numbers of people. Nonetheless, regulator-led action appears most likely, even if some markets are still setting up authorities. Weijdema expects the first case to be regulator led on paper, but very possibly complaint-led in reality, triggered by a consumer group, competitor, employee, journalist, civil society organization, or affected individual.
Unanswered accountability questions
Clients are repeatedly asking a question that has no good answer yet: How do we prove what an AI agent did, why it did it, and who was accountable?
In cybersecurity and governance, risk, and compliance, evidence matters. Logs, approvals, identities, access controls, retention, and audit trails are essential. But agentic AI can reason, retrieve data, generate content, and take actions across multiple systems. Governance must move from policy documents into technical controls.
Weijdema advises treating AI agents like privileged digital identities, with an owner, defined role, least-privilege access, monitoring, approval gates, and a kill switch. Organizations that get this right will not just be more compliant; they will be more resilient.
Another recurring question concerns where transparency ends and security testing begins. Security teams need realistic simulations, but the AI Act pushes toward disclosure when people interact with AI or are exposed to deepfakes. Designing exercises that remain realistic without crossing legal, ethical, or employee trust boundaries is the hard part.
Additional unresolved questions include who is ultimately accountable when an AI system causes harm—the vendor, deployer, business owner, or executive team—how to prove AI governance is working in practice rather than just documented in policy, and how much business value organizations are willing to lose to stay compliant, transparent, and auditable when using AI at scale.
As the first year of enforcement unfolds, organizations would be wise to focus on operational readiness, technical controls, and documentation. The regulatory landscape is still evolving, and the organizations that treat AI agents as privileged digital identities with clear accountability will be better positioned to navigate the uncertainties ahead.
Source: Help Net Security News