Colorado just passed a new AI law with more holes than Swiss cheese. After closed-door negotiations between Google, business groups, and consumer protection advocates, the tech industry walked away with nearly everything they wanted. The House passed SB 26-189 by a 57-6 vote on May 9, replacing Colorado's original 2024 AI law with something far more industry-friendly.

The carve-outs are stunning. Advertising, marketing, content moderation, and product recommendations are all explicitly excluded. Chatbots get a free pass if they have "acceptable use policies" (whatever those are). The bill only covers "consequential decisions" in narrow domains like employment, housing, lending, insurance, healthcare, education, and government services.

Here's what developers actually have to do: provide documentation. That's it. No requirement to disclose source code, model weights, or trade secrets. They need to keep records for three years and can't void contractual indemnification clauses with deployers. The law takes effect January 1, 2027, giving companies plenty of time to prepare their paperwork.

The timing creates a perfect contrast with Connecticut, which tightened its AI rules just days ago. Two states, same week, opposite directions. Connecticut said AI companies need to prove their systems are safe. Colorado said AI companies need to file some forms. Guess which state Google spent time lobbying in.

The bipartisan support is telling. Senate sponsors Robert Rodriguez and James Coleman, along with House sponsors Monica Duran and Jennifer Bacon, all signed off on this industry-friendly rewrite. When tech companies get Democrats and Republicans to agree on deregulation, you know the lobbying worked.

This is how AI regulation dies in America. Not with dramatic opposition, but with closed-door meetings, industry-friendly amendments, and lawmakers who think paperwork equals oversight. Colorado had a chance to lead on AI safety. Instead, they chose to follow Google's playbook.