The European Commission published draft non-binding guidelines on May 19 clarifying which AI systems qualify as high-risk under the Act. Teams shipping AI systems into the EU have until June 23 to submit feedback before those guidelines are finalized. August 2 is the hard date: after it, the full obligations of the Act apply, including conformity assessments, transparency requirements, and enforcement. This is distinct from the earlier advisory forum and scientific panel appointments (June 2026) — those were governance infrastructure. This is the compliance wall. If your team hasn't mapped your systems against the high-risk classification criteria, the June 23 window is the last structured opportunity to influence how that boundary is drawn.