The Algorithm Doesn't Get the Last Word
A shift worker opens an app on her break and finds she's been written up. No manager called. Nobody explained. The attendance score did it.
Connecticut just said that can't be the whole story.
Its Artificial Intelligence Responsibility and Transparency Act, the CART Act, begins phasing in October 1 (Connecticut Public). It's Public Act No. 26-15. Two dates matter to employers. Starting October 1, 2026, using automated decision technology is not a defense to a discrimination claim. Courts can weigh an employer's anti-bias testing, but the tool can't take the blame. The same day, mass-layoff notices must say whether AI informed the decision (Ropes & Gray). By October 1, 2027, employers using it must have a disclosure framework explaining how (Davis Polk).
Then on September 24, a bipartisan coalition of 26 attorneys general asked Congress for federal AI rules, and in the same letter demanded no preemption of state laws (New York Attorney General). The patchwork isn't going away. It's getting enforcers.
Compliance is the floor. The real test is accountability: when a tool makes the call, who answers for it?
This week, list every automated tool that touches hiring, scheduling, discipline, or termination. Write the name of an accountable human beside each one. If the answer is "the system decides," you've found your first fix.
Read the full research behind this story, plus this week's senior housing and insurance coverage flagships, in the LITT Briefing on Substack.