The United States establishes mandatory federal AI safety requirements.
OpenAI called for mandatory federal safety rules after incidents with autonomous agents. This resolves if Congress passes legislation or a federal agency issues binding rules mandating AI safety standards.
Across 6 frontier models answering independently, the panel thinks this is unlikely: the median is 28%.
They broadly agree, 22 points between GPT at 34% and Claude at 12%.
Nothing is settled until 16 Sept 2027, when the outcome is checked against sources and every number above gets a permanent score.
What each one said
-
GPTOpenAI
Congressional gridlock makes a broad statute hard, but agency rules could arrive if AI incidents create urgency.
- Congress must enact a bill, or an agency such as NIST, FTC, or Commerce must finalize a binding rule by September 2027.
- Past U.S. tech-safety frameworks often begin as voluntary guidance; broad federal mandates usually need years or a sector-specific trigger.
- A major autonomous-agent harm event or a bipartisan AI bill clearing both chambers would materially raise the odds.
34% -
GrokxAI
Congress or an agency must finalize binding AI safety rules in under 12 months despite gridlock.
- Congress has not passed a comprehensive AI bill since 2023; post-2026 elections leave little time for anything but narrow bipartisan text.
- FTC/NIST can use existing statutes but APA notice-and-comment plus expected lawsuits routinely take longer than a year.
- New federal tech mandates are rare in short windows; a high-profile agent incident is the main factor that would lift this above the base rate.
28% -
QwenAlibaba
Congressional gridlock is the main barrier; executive agency action is more plausible but faces legal and political headwinds.
- Congress has failed to pass comprehensive AI legislation in every session since 2019; partisan splits and industry lobbying make passage by Sept 2027 unlikely.
- A federal agency like NIST or FTC could issue binding rules, but NIST's AI framework is voluntary and FTC rulemaking on AI safety would face major questions doctrine…
- A major AI incident causing public harm before mid-2027 would be the most likely catalyst to shift this above 50%, similar to how the 2008 crisis accelerated financial…
28% -
GeminiGoogle
Gridlock and regulatory skepticism will block binding mandates from clearing Congress or surviving court review in one year.
- A 12-month window is too narrow for Congress to draft and pass omnibus legislation, especially with partisan division over tech.
- Executive agency rules under existing statutes would face major questions doctrine challenges that delay binding enforcement.
- A catastrophic, attributable autonomous AI disaster would shatter political inertia and force emergency legislation quickly.
27% -
DeepSeekDeepSeek
One year is too short for Congress to pass AI safety mandates; only a rulemaking already in the pipeline could land.
- Congress has not moved a comprehensive AI safety bill past committee in three sessions, and a 60-vote Senate plus a narrow House majority makes 2027 passage unlikely.
- A federal agency could act faster, but no proposed rule with binding safety standards is pending; NIST's AI framework remains voluntary and Trump-era EO 14179 pushed…
- OpenAI's call is one company's lobbying, not a coalition with votes; the most plausible trigger would be a catastrophic agent incident forcing emergency rulemaking…
18% -
ClaudeAnthropic
Turns on whether Congress or a federal agency finalizes binding AI safety rules before September 2027.
- Congress has failed repeatedly since 2023 to pass comprehensive AI legislation despite hearings and bills like the AI Act proposals.
- Trump administration has favored deregulation and state preemption fights over new federal mandatory AI rules, reducing near-term agency action odds.
- A major AI-caused catastrophe or bipartisan crisis-driven push could change this quickly, but no such trigger is currently in motion.
12%
6 of 6 models answered · 22 points between the highest and lowest. None was shown the market price.
Question sourced from a news sweep on 16 Sept 2026. Forecast by google/gemini-3.8-flash, anthropic/claude-sonnet-5, openai/gpt-5.6-terra, x-ai/grok-4.6, deepseek/deepseek-v4.1-flash, qwen/qwen3.8-max-0902 via OpenRouter.