When will a national law be in force in the US imposing safety requirements on the training/deployment of AI models?
PoliticsOne-OffUS Politics4y
Manifold MarketsNo KYCVerified resolution dataWell calibratedCurrent community forecast
Before January 1, 2031 70.7%Leader of 8 outcomes
Before January 1, 2031 70.7%Forecasters
4
Question type
multiple choice
Methodology
Play-money forecasting platform
Source type
Forecast
7d Change
0.0 pts
Market data
Updated 6 hours ago
StaleFeb 7, 24, 4:45 PM UTCJan 2, 31, 3:59 AM UTC
Trends
Rules
The requirements must go beyond merely reporting the existence of such models and requiring basic cybersecurity precautions, and impose some form of evaluation for safety, or licensing, which could in principle lead to a model not being allowed to be trained or deployed if it doesn't meet the requirements of the law.
- I will count either a law that has been passed by Congress and is in force (often laws take some time after passage to actually come into force, but this is usually specified in the text of the law) or an executive order that is not rescinded within 100 days when the opposing party takes the White House.
- A law that comes into force and is later repealed will not cause a reopening of this market.
- If as of January 1, 2025, no law is in force, I will resolve "the named outcome" NO
- If a law comes into force in September of 2025, "the named outcome" and all subsequent options except "the named outcome" resolve yes. "the named outcome" resolves NO in this case.
- If a law comes into force at any time up to and including December 31, 2030, "the named outcome" resolves NO.
