Connecticut's House passed Senate Bill 5 — now formally called the Connecticut Artificial Intelligence Responsibility and Transparency Act — by a 131-17 vote, with the Senate having passed it 32-4; Governor Lamont has said he plans to sign it.

The 71-page omnibus bill covers AI in employment decision-making, frontier model whistleblower protections, chatbot safety rules (including self-harm detection protocols), and AI provenance/disclosure requirements.

On employment, developers of AI tools used as a "substantial factor" in hiring, promotion, discipline, or discharge must supply compliance information to deployers, and the bill codifies that automated decision-making is not a defense to a discrimination claim — going further than any other state.

Frontier model whistleblower protections kick in October 1, 2026: developers training models above 10²⁶ compute operations must protect employees who report catastrophic-risk concerns, with large developers (over $500M revenue) required to establish anonymous internal reporting by January 1, 2027.

Enforcement rests primarily with Attorney General William Tong, who previewed an aggressive posture in a February 2026 advisory memo; the bill also leaves open a private right of action for its youth-protection provisions.

The Trump administration has pushed back on state-level AI laws, arguing for a single federal standard — making Connecticut's move a direct defiance of that pressure and a bellwether for what's coming in other states.

Colorado Senate Majority Leader Robert Rodriguez filed a bill to repeal and replace the Colorado AI Act — the nation's first comprehensive AI law — with a far lighter disclosure-based regime that strips out most of its key compliance mandates.

The replacement bill is the product of Governor Polis's working group and removes core requirements around algorithmic discrimination, risk management programs, and impact assessments that had put significant compliance burdens on AI developers and deployers.

Colorado's legislature closes May 13, meaning the repeal-and-replace vote is imminent; if it passes, a landmark law that had been scheduled to take effect June 30, 2026 would effectively be gutted before it ever went live.

Chatbot legislation is also surging across the country: bills advanced in Oklahoma, Hawaii, Michigan, and New York this past week, with Oklahoma and Hawaii appearing poised for passage in the near term.

Maryland became the first state to ban certain AI-driven price-setting practices, signing HB 895 into law — a direct signal that algorithmic pricing is the next major frontier for state AI enforcement.

Founders building AI products in any U.S. state should treat this week's activity as a five-alarm warning: the patchwork is accelerating fast, with different definitions, standards, and mandates proliferating state by state.

At today's Code w/ Claude 2026 developer event, Anthropic announced it is partnering with SpaceX to use the full capacity of SpaceX's Colossus data center in Memphis — a massive infrastructure commitment signaling Anthropic's aggressive push to dominate enterprise agentic AI.

Anthropic's Chief Product Officer Ami Vora led the keynote with no new model announcement, instead focusing on platform-level improvements: updates to Claude managed agents, multi-agent orchestration, and a new "Claude Code routines" feature for developers.

Rate limits for Claude Code are being doubled for Pro, Max, and Enterprise customers effective immediately, and API traffic has grown 17x since last year — a signal of the scale at which developers are now building on the Claude platform.

The "advisor strategy" was highlighted as a key architectural pattern: Sonnet calling Opus as an on-demand advisor achieves frontier-model quality benchmarks at 5x lower cost, a compelling efficiency play for startups building agentic products on Claude.

Major enterprise customers Shopify and Mercado Libre were cited; Mercado Libre (with 23,000 engineers) is targeting 90% autonomous coding by Q3 2026 — illustrating how fast AI-native development workflows are becoming table stakes at scale.

For SaaS founders, the Colossus deal and doubling of rate limits are the clearest signs yet that Anthropic is positioning its platform — not just its models — as the critical infrastructure layer for the agentic AI era.