During the AI Act trilogue: confine foundation-model/GPAI obligations to transparency, keep the risk-based approach targeting high-risk uses rather than technologies, and reject 'very capable'/'high-impact' model tiers, the industry counter-position to comprehensive GPAI regulation.
Pressed by DIGITALEUROPE and 30+ national associations at trilogue stage (Nov 2023). The December 2023 political agreement and final AI Act retained tiered GPAI obligations, including systemic-risk duties beyond transparency for models above the compute threshold. The ask's successors live on in Code of Practice scope fights and omnibus simplification demands.
first seen 2023-11-23 · last seen 2023-12-08 · locus: developer
Bill-status would miss this
1 claims · every quote links to its document
“It is vital that we reduce this burden as much as possible, and let SMEs apply GPAI, foundation models and other new emerging AI technologies in their innovations. This is why we support recent moves by Member States to limit the scope for foundation models to transparency standards.”
Joint statement with 30+ national digital industry associations, at AI Act trilogue stage (Nov 2023).
No on-the-record claims captured on this side. Absence here is a coverage fact, not a finding.