California SB-1047: Safe and Secure Innovation for Frontier AI Models Act — vendor disclosure form

1. Vendor identification

2. AI system identification

3. Provision-specific questions

1. Foundation Models / GPAI. Does the offered system meet the threshold for a general-purpose / foundation model under California SB-1047: Safe and Secure Innovation for Frontier AI Models Act (Cal. SB-1047 §22602 — 'covered model' = trained with >10^26 operations AND >$100M cost (or fine-tuning >$10M); vetoed 29 Sep 2024)? If yes, identify the specific obligations you will satisfy and the evidence you will provide.

   (Cite: Cal. SB-1047 §22602 — 'covered model' = trained with >10^26 operations AND >$100M cost (or fine-tuning >$10M); vetoed 29 Sep 2024)

2. Compute-Threshold Reporting. Does the offered system meet the compute / training-cost reporting thresholds named in California SB-1047: Safe and Secure Innovation for Frontier AI Models Act (Cal. SB-1047 §22603(b) — annual reporting of training compute + safety determination)? If yes, attach the most recent submitted report (redacted as needed) or confirm submission status.

   (Cite: Cal. SB-1047 §22603(b) — annual reporting of training compute + safety determination)

3. Catastrophic & Existential Risk. Has the offered system been evaluated against catastrophic-risk thresholds (e.g., CBRN information uplift, autonomous replication) consistent with California SB-1047: Safe and Secure Innovation for Frontier AI Models Act (Cal. SB-1047 §22602 — defines 'critical harm' including mass casualties, $500M+ damage)? Provide the evaluation report or its public-disclosure equivalent.

   (Cite: Cal. SB-1047 §22602 — defines 'critical harm' including mass casualties, $500M+ damage)

4. Open-Weight Frontier Release. Are the model weights powering the offered system released under an open-weight licence? If yes, identify the licence + any restrictions under California SB-1047: Safe and Secure Innovation for Frontier AI Models Act (Vetoed bill — would have required covered models (incl. open-weight releases) to adopt a safety & security protocol + self-certified compliance, with independent third-party audits from 2026 (Anthropic + Meta objected on different grounds)).

   (Cite: Vetoed bill — would have required covered models (incl. open-weight releases) to adopt a safety & security protocol + self-certified compliance, with independent third-party audits from 2026 (Anthropic + Meta objected on different grounds))

4. Documentation enclosures expected

Tick each enclosure attached to the vendor response. Missing enclosures should be explained in the “Variances” field below.

• Copies of submitted regulatory reports / registrations
• Safety / capability evaluation results
• End-user redress + complaint-channel procedure
• Vendor company registration + insurance certificates
• Sub-processor / supply-chain list (including model upstream)

5. Vendor attestation

The undersigned, on behalf of the vendor, attests that the disclosures above are true and complete to the best of their knowledge at the date signed, and undertakes to notify the buyer in writing within 30 days of any material change to those disclosures.