The EU Commission published draft guidelines on Article 6 high-risk classification on May 19, 2026. The consultation window is open until June 23, 2026. Most compliance coverage has focused on the deadline dates and the technical criteria. It missed something more operationally disruptive.
According to legal analysis from Wilson Sonsini, LEXR, and Alston & Bird published in the days following the guidelines’ release, a provider’s marketing language, sales materials, and technical documentation are active legal factors in determining whether an AI system falls under the high-risk classification, not only the system’s underlying technical function. The EU AI Act’s classification framework, per Article 6(1), considers the system’s intended purpose. That intended purpose is established, in part, by what the provider says the system does. Your go-to-market copy is evidence.
That’s a consequence most organizations aren’t structured to handle.
Compliance programs built around technical architecture reviews haven’t touched sales decks or pre-sales qualification scripts. Product marketing teams writing about AI capabilities haven’t been briefed on how their language maps to Annex III risk categories. According to LEXR’s analysis, the gap between what a system technically does and how it’s marketed is exactly where classification ambiguity lives, and where regulators will look first when a dispute arises.
EU AI Act: What Determines High-Risk Classification
Two deadlines anchor the compliance calendar. Providers and deployers of standalone high-risk AI systems under Annex III, covering recruitment, biometrics, credit scoring, education, and law enforcement applications – must meet compliance obligations by December 2, 2027. Providers of AI embedded in regulated products under Annex I (machinery, medical devices, toys) face a later deadline of August 2, 2028. The marketing-as-trigger dynamic applies to both tracks: if your documentation and sales materials describe your system as operating in an Annex III sector, that description contributes to classification.
The consultation deadline matters for practitioners with ambiguous products. The June 23 window is the formal opportunity to submit comments to the EU Commission before the guidelines are finalized. Legal analysts note the draft guidelines are likely to carry significant weight with national regulators and courts even before finalization, Commission guidance of this type has historically been treated as persuasive authority. Don’t expect finalization to reset the clock on internal classification reviews.
The catch is that penalties make passive compliance expensive. Severe EU AI Act violations carry fines of up to €35 million or 7% of global annual turnover under the Act’s enforcement provisions.
For most organizations, the near-term action isn’t a documentation overhaul. It’s a classification audit that includes the marketing team. Pull the product’s sales materials, pre-sales qualification scripts, and technical data sheets. Map them against the Annex III categories. If your go-to-market copy describes capabilities in sectors covered by Annex III, you’re building a classification record, intentionally or not.
Who This Affects
According to Alston & Bird’s analysis via JDSupra, the draft guidelines represent the Commission’s clearest signal yet that classification isn’t purely an engineering question. The organizations that treat it as one will be the least prepared when national regulators begin enforcement.
The real question is whether your legal team knows what your sales team is saying. Most don’t. That asymmetry is what the draft guidelines just made expensive.