Washington is still improvising on AI. Industry warnings have grown louder as lawmakers struggle to decide whether artificial intelligence needs guardrails, permission slips, or another study group with better catering.
Jacob Coxon, a former employee at OpenAI and Anthropic, warned that AI companies were “gambling with our lives” by building powerful models that could destroy humanity. His post drew over 100 million views within days, and Evan Hubinger, an Anthropic researcher, publicly backed the claim while putting greater than 10% odds on human extinction within a decade.
That warning landed on top of a policy process that began years before ChatGPT. Congress introduced the FUTURE of AI Act in 2017 to create a federal framework for studying AI, then folded the National Artificial Intelligence Initiative Act into the defense bill in 2020 to fund research and workforce training.
The industry accelerated after ChatGPT launched in November 2022. On May 16, 2023, OpenAI CEO Sam Altman testified before the Senate Judiciary Committee, asked the government to regulate AI, and proposed a licensing agency. One year later, requiring government approval would become “disastrous” in Altman’s own words on May 8, 2025. Regulation has a way of looking better before it reaches the paperwork stage.
Congress keeps studying the problem
In June 2023, Senate Majority Leader Chuck Schumer announced nine closed-door “AI Insight Forums” for tech CEOs. The House Science Committee approved nine bipartisan AI bills in September 2023, but those measures focused on research and education rather than safety.
The gap between technical risk and legislative understanding surfaced in plain language. Representative Sam Liccardo said, “There is a percentage of my colleagues who are still trying to spell AI.” Representative Chip Roy offered a broader warning: “You cannot just put your head in the sand on AI and say, ‘I hope it works out.’ You can’t just go in and say, ‘Government’s got the solution — they’re just going to ban it all.’ That’s just not going to work.”
State lawmakers moved faster, though federal politics kept trying to pull the emergency brake. California passed SB 1047 in August 2024, requiring safety testing for the largest AI models, but Governor Gavin Newsom vetoed it in September 2024. President Donald Trump revoked President Joe Biden’s executive order on AI on his first day in office and ordered an “AI Action Plan.”
Senator Ted Cruz, R-Texas, later inserted language into the “One Big Beautiful Bill” that would have barred states from enforcing AI laws for 10 years. The Senate removed that language by vote on July 1, 2025, leaving states with room to write their own rules.
The AI fight moved from policy to political warfare
California enacted SB 53 in September 2025, and New York’s RAISE Act became law. OpenAI funded a super PAC that spent more than $7.6 million opposing Assemblymember Alex Bores, sponsor of the RAISE Act. Roughly $15 million to $19 million supported Bores, while outside money in the race topped $20 million.
In June 2026, Bores lost the primary despite that spending. The result showed the limits of corporate influence, or at least the limits of buying the outcome in a race where everyone arrived with a very large checkbook.
Trump signed an executive order on December 11, 2025, creating a Justice Department “AI Litigation Task Force” to sue states over AI laws. In April 2026, Colorado’s law was gutted after lawsuits from xAI and OpenAI, turning state-level safety rules into targets for courtroom campaigns.
By September 2026, disagreements over AI safety legislation had reached the Senate again. Senator Maria Cantwell, D-Washington, opposed a bill that preempted state laws and said she “welcomes the urgency” to get something done, but the “answer is not a weak federal standard that becomes a backdoor for wiping out stronger state protections.”
The political outlook is not subtle: “Disagreements over how far to go on AI safety legislation portend a rocky road ahead for bipartisan Senate talks on a bill that remains the most likely vehicle for action this year.” Lawmakers face a technology moving faster than their process, companies that want access to government when it suits them, and resistance when oversight becomes real.
That leaves the United States with competing instincts rather than a stable policy. Congress wants research, workforce training, and national leadership; states want safety rules; AI companies want room to build. The result, as of September 12, 2026, is a regulatory contest where the models keep advancing and the lawmakers keep arguing over who gets to hold the leash.
Based on
- Aging Congress Is Completely Ill-Equipped to Regulate AI, or Seemingly Even Understand It — futurism.com
- After a decade of failed bills and resignations, Washington discovers it cares about AI safety | Fortune — fortune.com
- AI’s existential threat suddenly has Washington’s attention – The Washington Post — washingtonpost.com
- Disagreements reemerge in bipartisan AI safety talks | Semafor — semafor.com
- More US lawmakers seek new AI rules after researchers warn of human extinction – France 24 — france24.com
