What happened
As of mid-July 2026, providers of general-purpose AI models were finalizing compliance ahead of August 2, when the EU AI Office gains full enforcement powers under the AI Act. From that date regulators can request information, order model recalls, mandate mitigations, and impose fines, backed by a finalized Code of Practice covering transparency, copyright, and safety. Obligations themselves have applied since August 2025, but the enforcement teeth arrive now.
The context
It is the first time a major jurisdiction can meaningfully penalize the makers of the largest AI models.
Why this moves the Board: Signal only · T3 Reported · context for the field, not a direct score input.