x.com/DarioAmodei/status/2088758816376807762
1 correction found
completely exempt any company below a certain amount of revenue or model training costs from being covered at all (it was $500M for SB 53, lower for 1047 but we objected to that).
SB 53 does not fully exempt sub-$500M companies. Smaller frontier developers are still covered by some SB 53 requirements, including public transparency-report duties.
Full reasoning
This overstates what California's SB 53 does.
SB 53 defines a "large frontier developer" as a frontier developer with more than $500 million in annual gross revenue. That threshold matters for the bill's additional obligations on large developers, such as publishing a full frontier AI framework.
But the statute also separately regulates any "frontier developer" that trains or deploys a frontier model above the compute threshold. In particular:
- Section 22757.12(c)(1) says that "a frontier developer" must publish a transparency report before or when deploying a new or substantially modified frontier model.
- Section 22757.12(e)(1)(A) says "a frontier developer shall not make a materially false or misleading statement" about catastrophic risk from its frontier models or its management of that risk.
So a company that is below $500M in revenue is not "completely exempt" from SB 53 if it is still a frontier developer under the bill's compute threshold. The $500M cutoff determines who counts as a large frontier developer, not whether smaller frontier developers are "covered at all."
This correction is about the SB 53 portion of the sentence; I am not disputing here that SB 1047 used lower compute-cost thresholds.
1 source
- Compare Versions - SB-53 Artificial intelligence models: large developers
"Large frontier developer" means a frontier developer ... with annual gross revenues in excess of five hundred million dollars ($500,000,000)...; "a frontier developer shall clearly and conspicuously publish ... a transparency report" before deploying a new frontier model; and "a frontier developer shall not make a materially false or misleading statement" about catastrophic risk from its frontier models.