Courts, law firms and newsrooms aren’t afraid of AI. They run on a chain of accountability that runs back to a human being, and generative text cuts it.
New Hampshire’s 2024 AI child-abuse-image clause needs no real child and no obscenity finding. The state Supreme Court rejected that reasoning in 2008 in State v. Zidel, the U.S. Supreme Court rejected it in 2002, and on Aug. 25 the Seventh Circuit refused to enforce the narrower federal version. The committee that wrote the clause called the missing element a loophole.
New Hampshire did not wait for Congress: three 2024 laws criminalize deepfakes made to harass, defame, or extort, require disclosure of deceptive AI in political ads, and set the first statewide ground rules for AI used by state agencies, months after fake Biden robocalls hit the primary.
The U.S. is creating a complex regulatory framework for AI, marked by Executive Order 14110, agency enforcement, and state-level innovations. This evolving system combines voluntary standards and existing laws to manage risks, ensure accountability, and protect civil rights. The focus will shift from sweeping legislation to a standards-led, enforcement-driven approach.
The development of ethical AI is the responsibility of human programmers, encompassing a lifecycle approach from problem framing to monitoring incidents. Ethical considerations must guide data sourcing, modeling, and evaluations. Compliance with regulatory frameworks like the EU AI Act and NIST guidelines is essential to managing risks and ensuring public trust in AI products.
