March delivered a collision between state action and federal ambition on AI regulation. Washington and Oregon signed the first chatbot safety laws in the country, Virginia passed four AI bills, and the White House released a legislative framework explicitly aimed at preempting state laws. Here’s what’s moving.
Washington Signs Five AI Bills Into Law
Governor Bob Ferguson signed five AI-related bills before Washington’s legislative session ended March 12, making the state the most active on AI regulation this spring.
HB 2225 - Chatbot Safety for Minors
The flagship bill, a personal priority for Ferguson, requires companion chatbot operators to:
- Implement protocols for detecting and responding to expressions of self-harm or suicidal ideation
- Refer users in crisis to appropriate resources
- Disclose that the chatbot is AI at the start of interaction and every three hours thereafter
- Prohibit generating content that encourages self-harm
The bill passed the House 69-28 and the Senate 43-5. It takes effect January 1, 2027.
Other Washington Bills:
- HB 1170 - Requires disclosure when content is AI-generated or AI-modified
- SB 5395 - Regulates AI use in health insurance decisions
- SB 5105 - Addresses AI-generated deepfakes and sexually explicit depictions of minors
- SB 5886 - Establishes digital likeness property rights
Washington’s HB 2225 includes a private right of action, meaning individuals can sue chatbot operators directly for violations without waiting for the state to act.
Oregon: First Chatbot Law With Real Teeth
Oregon’s SB 1546 passed the Senate 26-1 and House 52-0 on March 5, making it the first chatbot safety bill to pass a legislature this session.
The bill requires chatbot operators to:
- State upfront that the chatbot may not be suitable for minors
- Remind minor users they’re interacting with AI, not a human, at least once per hour
- Implement self-harm detection and response protocols
- Report annually to the Oregon Health Authority on crisis intervention incidents
Legal analysts are calling SB 1546 the first chatbot law with real teeth. Like Washington’s bill, it includes a private right of action and takes effect January 1, 2027.
Governor Tina Kotek has 30 weekdays to sign, veto, or let it become law without her signature.
Virginia Passes Four AI Bills
Virginia’s session ended March 14 with four AI bills signed into law:
- HB 580 - AI fraud and abuse protections
- HB 797 - Framework for independent verification organizations
- SB 245 - Duty of care for social media platforms regarding minors
- HB 1186 - Prohibits mandatory student interaction with AI chatbots in schools
The student chatbot bill is notable. Schools can still offer AI tools, but they cannot require students to use them or penalize students who decline.
Vermont: AI Election Deepfake Disclosure
Governor Phil Scott signed S.23 on March 5, requiring disclosure of AI-generated content in campaign materials used within 90 days of an election.
The law mandates visible, easy-to-read labels on any AI-generated images, audio, or video. It builds on Vermont’s 2024 law criminalizing non-consensual AI-generated intimate images.
Florida’s AI Bill of Rights Dies
Governor DeSantis’s “AI Bill of Rights” (SB 482) passed the Florida Senate but failed to advance in the House before the session ended March 13. The bill would have established consumer rights around AI decision-making but faced industry opposition over compliance costs.
White House Releases Federal AI Framework
On March 20, the Trump administration released a legislative framework for AI that explicitly targets state regulation.
Six Guiding Principles:
- Protect children from AI harms
- Prevent electricity costs from surging due to AI data centers
- Respect intellectual property rights
- Prevent algorithmic censorship
- Educate Americans on AI use
- Remove barriers to innovation
The framework calls for federal preemption of state AI laws, arguing that “a patchwork of conflicting state laws would undermine American innovation and the nation’s ability to lead in the global AI race.”
Industry Response:
The AI industry praised the framework, with lobbying groups saying it shows the White House understands that a “light-touch regulatory environment” is required for innovation.
State Pushback:
California Governor Gavin Newsom’s office criticized the framework, saying Trump is “trying to gut laws in California that keep residents safe.” Colorado and Utah officials have also pushed back, noting their AI laws are already in effect.
Republican Rebellion:
The framework faces opposition from within Trump’s own party. On March 3, more than 50 Republican state legislators from 22 states sent a letter urging the president to ease federal pressure on state AI legislation.
The letter followed a February 15 White House memorandum opposing Utah’s HB 286, which the administration called “an unfixable bill that goes against the Administration’s AI Agenda.” The bill would impose transparency requirements and youth safeguards on large AI developers.
The GOP lawmakers wrote they are “deeply concerned by the work of officials seeking to pressure lawmakers in Utah and other states to abandon legislation aimed at mitigating risks at leading AI labs and safeguarding constituents, including young people, from AI’s worst harms.”
They argued the administration’s efforts “risk infringing on states’ rights and undermining a core constitutional principle that conservatives have long defended.”
EU Moves to Simplify AI Act
While the US debates whether to have any AI regulation at all, the EU is streamlining rules it’s already passed.
On March 13, the Council of the European Union adopted its negotiating position on amendments to the AI Act as part of the “Omnibus VII” simplification package.
Key Proposed Changes:
- Delay enforcement of high-risk AI system rules by up to 16 months, until standards and technical tools are confirmed ready
- Extend regulatory exemptions currently available to SMEs to small mid-caps
- Reduce certain documentation requirements in limited cases
- Allow more processing of sensitive personal data for bias detection and mitigation
- Strengthen the AI Office’s powers while reducing fragmentation across member states
The Council position is a mandate for trilogue negotiations with the European Parliament and Commission. The timeline targets a vote in June and publication of amendments in July 2026.
This is simplification, not weakening. The AI Act remains fully operative—the changes aim to make compliance more practical while maintaining core protections.
Colorado AI Act Still Delayed
Colorado’s landmark AI discrimination law remains in limbo. Originally set to take effect February 1, 2026, the compliance deadline is now June 30, 2026.
Governor Jared Polis signed the delay bill (SB 25B-004) in August 2025 after negotiations on substantive amendments collapsed. The law’s requirements—impact assessments for high-risk AI systems, disclosure obligations, and discrimination prevention measures—remain unchanged.
State legislators from both parties have promised to resume negotiations when the 2026 session begins in January. Whether they can reach compromise before the June deadline remains uncertain.
What’s Still Moving
Georgia (session ends April 6):
- SB 540 - Chatbot safety provisions advancing
- SR 789 - AI impact study resolution
- SB 444 - AI in healthcare insurance decisions
New York:
- GenAI warning bill passed legislature, awaiting Governor Hochul
- RAISE Act chapter amendment passed both chambers
Utah:
- HB 276 (Digital Content Provenance Standards Act) passed legislature
Hawaii:
- Surveillance pricing prohibition in food sales passed House
What This Means
The collision course between state and federal AI regulation is now explicit.
States aren’t waiting. Washington signed five bills. Oregon and Vermont signed chatbot and deepfake laws. Virginia passed four measures. These laws take effect in 2027 regardless of what happens in Congress.
The White House framework makes federal preemption a stated goal. But executive frameworks don’t override state law—that requires Congressional action. And Congress hasn’t passed comprehensive AI legislation yet.
For companies building AI chatbots, the immediate compliance obligation is clear: Washington and Oregon’s laws require self-harm detection, crisis referral protocols, and disclosure requirements by January 2027. The private right of action means litigation risk, not just regulatory enforcement.
For states, the federal framework is a warning shot but not a legal constraint. California, Colorado, and other states with existing AI laws are signaling they won’t back down.
The question for 2026 isn’t whether AI will be regulated—it’s who gets to write the rules. States are answering that question while Washington debates.