The Responsible AI Safety and Education Act (RAISE Act) is a New York AI safety bill that requires AI developers to produce safety protocols to address the most severe risks (and then follow those protocols), and creates whistleblower protections for AI company employees.
This market will resolve based on the furthest stage the bill reaches by January 1, 2026. Possible outcomes include:
Not approved by the legislature
Approved by the legislature but did not become law (e.g. because the governor vetoed it)
Has become law
Substantial amendments to the bill (e.g. ones that significantly water it down) will not change the resolution of this market, unless the amendments are so substantial that it's a fundamentally different bill (as happened with TRAIGA in Texas). Sorry that this isn't a clear-cut resolution criterion -- I'll make judgments on a case-by-case basis, though I suspect that it won't end up being relevant.
Federal preemption is not relevant to the resolution of this market. For instance, if the governor signs the bill but federal preemption prevents it from taking effect, the market nevertheless resolves "Has become law".
Update 2025-12-06 (PST) (AI summary of creator comment): Regarding amendments to the RAISE Act:
Substantially weakened versions of the bill will still resolve to the appropriate outcome (e.g., "Has become law" if signed)
A NO resolution would only occur if the bill becomes fundamentally different (a complete change to what the bill does, not just a weakened version)
There remains inherent ambiguity around the exact boundary for extreme weakening cases
Update 2025-12-11 (PST) (AI summary of creator comment): If the RAISE Act is replaced by SB 53 near-verbatim, the market will resolve to "Not approved by the legislature" (NO).
1,000I just want to clarify this part of the resolution criteria:
Substantial amendments to the bill (e.g. ones that significantly water it down) will not change the resolution of this market, unless the amendments are so substantial that it's a fundamentally different bill (as happened with TRAIGA in Texas).
If the RAISE Act is really substantially weakened, this still resolves YES. A NO resolution would entail a complete change to what the bill does, as if it was a totally different bill, rather than a weakened version of the same bill. There's still an ambiguity around cases where the bill is extremely weakened (if the final version of the bill is literally empty, this will resolve NO), but there will inevitably be ambiguity around where the boundary is exactly, sorry.