Resolves YES if the European AI Office or an EU member state authority publicly announces a formal enforcement action under the EU AI Act (Regulation 2024/1689) against a provider of a General-Purpose AI (GPAI) model with systemic risk (e.g. OpenAI, Anthropic, Google DeepMind, Meta, Mistral, xAI).
Qualifying "formal enforcement actions" must be publicly announced and include any of the following:
A formal investigation
A formal Request for Information (RFI)
A notice of non-compliance or fine
A binding order to withdraw, modify, or correct an AI system
Exclusions: Informal guidance, warnings without legal force, or voluntary commitments do not count. The ultimate success or conclusion of the action is irrelevant for resolution.
There will be no AI clarifications added to this market's description. Resolution = expiry.
1,000Which of these would count as frontier, given Mistral seems to count per the description?
Alibaba
Baidu
DeepSeek
Moonshot
@PlasmaPower Excellent question my love 💐👒
As I understand it, to trigger an enforcement action, a Chinese AI company must meet two legal requirements under the Act:
The Capability Threshold: The model must be trained using more than 10^25 FLOPs of compute, OR the EU Commission must manually designate it as having "systemic risk" due to its high-impact capabilities.
The Jurisdiction Threshold: The model must be "placed on the market or put into service" within the European Union.
To my knowledge this includes only Alibaba and DeepSeek from that list. Though note that any company counts as long as the EU AI Office officially classifies their model as a GPAI 🌷