Direct comparison
Seoul Pledges vs SB 53/RAISE Act: Enforceable?
The Seoul AI Safety Commitments are voluntary; SB 53 and the RAISE Act are statutes with AG enforcement and civil penalties. What that means in practice.
Written and maintained by CASRAI Editorial Board
Last updated
Ask CASRAI · free to try
Ask about Seoul Pledges vs SB 53/RAISE Act: Enforceable?
Ask your first 2 questions free below. Subscribers get 150 a day for $29 a month.
Ask CASRAI answers research-administration questions and cites the passages behind every claim. When our sources don't cover a question, it says so.
Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.
Works on this site and inside Claude, Cursor and the AI tools you already use.
Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays free to read, with no account and no card.
How do Seoul Frontier AI Safety Commitments, SB 53 / RAISE Act (state statute) compare side by side?
The table below compares Seoul Frontier AI Safety Commitments, SB 53 / RAISE Act (state statute) across 4 procurement-relevant dimensions, from legal status through real-world track record.
Side-by-side comparison
| Dimension | Seoul Frontier AI Safety Commitments | SB 53 / RAISE Act (state statute) |
|---|---|---|
| Legal status | A voluntary multilateral political pledge, not a law. Signed by 20 AI developers at or after the May 2024 Seoul AI Safety Summit (16 original signatories, plus 01.AI, Magic, MiniMax, and NVIDIA added in February 2025). No statute, regulator, or court gives the pledge legal force. | Codified state law in two states. California SB 53 (the Transparency in Frontier Artificial Intelligence Act) took effect January 1, 2026. New York's RAISE Act (S6953B, signed by Governor Hochul December 19, 2025 as Chapter 699 of the Laws of 2025) took effect on the 90th day after signing, per the bill's own effective-date clause — both are in force as of this writing. |
| Enforcement mechanism | None. There is no regulator, complaint process, or filing requirement attached to the pledge. The only pressure is reputational: each signatory is expected to publish its own safety framework, and outside observers can compare that framework against the commitment text. | State Attorney General only. SB 53 is enforced exclusively by the California Attorney General through civil actions; the RAISE Act is enforced by the New York Attorney General on the same basis. Neither statute creates a private right of action — an individual or advocacy group cannot sue a developer directly under either law. |
| What happens if a company doesn't comply | Nothing legally. CASRAI's Seoul Commitments guide is explicit that "they are not a law — no company can be fined for missing a commitment." A signatory that never publishes a safety framework, or publishes a thin one, faces scrutiny from press and civil-society trackers, but no fine, injunction, or compliance order can follow. | Civil penalties, brought by the state. Under SB 53, penalties run up to $1,000,000 per violation, scaled to severity. Under the RAISE Act as enacted, a first violation carries penalties up to $10,000,000 and subsequent violations up to $30,000,000; the statute also authorizes injunctive and declaratory relief. Both routes run through the Attorney General's office, not a private lawsuit. |
| Real-world track record | Mixed and largely undocumented. There is no central registry confirming which signatories have met the pledge's terms; CASRAI's tracking notes compliance "varies by company," and no signatory has faced any consequence — formal or informal — for falling short, because none is possible under the pledge itself. | Too early to assess. SB 53's core duties became binding January 1, 2026 and the RAISE Act's roughly 90 days after its December 2025 signing — both within the past year. As of September 2026, CASRAI is not aware of a completed, publicly reported enforcement action by either the California or the New York Attorney General under these provisions. |
Common questions
Common questions about Seoul Frontier AI Safety Commitments vs SB 53 / RAISE Act (state statute)
Are the Seoul Commitments legally binding?
+
No. They are a voluntary political pledge signed at the Seoul AI Safety Summit by 20 AI developers. No statute enforces them, and no company can be fined, sued, or sanctioned for failing to meet one. CASRAI's own NIKOLAI project — an independent, unendorsed reference dictionary of frontier-AI-safety terms, not an assertion that any Seoul signatory has adopted it — maps exactly this kind of voluntary pledge as its Commitment element, in the Commitments and Governance track (<a href="https://casrai.org/nikolai/track/n9-commitments-governance">nikolai/track/n9-commitments-governance</a>), distinct from the statutory duties SB 53 and the RAISE Act impose.
Are SB 53 and the RAISE Act both actually in force right now?
+
Yes, as of this writing. SB 53's core obligations took effect January 1, 2026. The RAISE Act was signed December 19, 2025 and took effect on the 90th day after signing under its own effective-date clause, roughly mid-March 2026.
Can a company be a Seoul Commitments signatory and also be covered by SB 53 or the RAISE Act?
+
Yes. Several original Seoul signatories — including Google, Meta, Microsoft, and OpenAI — also meet SB 53's and/or the RAISE Act's compute- and revenue-based thresholds for a "large frontier developer." A company can hold both the pledge and the statutory duty at once; only the statutory duty is enforceable.
Has any company actually been fined under SB 53 or the RAISE Act?
+
Not that CASRAI is aware of as of September 2026. Both statutes' core duties only became binding within the past year, and neither state's Attorney General has publicly reported a completed enforcement action under these provisions yet.
Going deeper
