Terms for a model that changes: GSA’s AI contract clause takes effect
GSA’s clause for buying large language models took effect October 19. Its change notice and suspension right treat the model as something that changes after award.
GSA’s clause for buying large language models took effect October 19. Its change notice and suspension right treat the model as something that changes after award.
Character.AI filed for bankruptcy, and 40 attorneys general moved to protect users’ chat data in any sale. The question for buyers is what their own contracts say.
An AI pilot can look successful right up until renewal. Denver’s rejected chatbot contract shows why AI governance needs evidence about accuracy, risk, cost, and performance, not just usage and satisfaction.
Plano approved up to $390,000 for citywide AI, and $150,000 of it pays for a tool that monitors how the city uses AI. The oversight came in the same purchase.
Raleigh gave its AI help desk agent new-hire permissions and pulled it back when it overreached. The vendor’s 66 percent savings figure has no published baseline.
Colorado’s rewritten AI law takes effect in January. It trades impact assessments for notices and explanations, and the draft rules decide how specific those must be.
Casper’s council moved a police AI tool from trial to a $58,800 contract on two consent votes. The pilot’s measures were never published, so the renewal has no baseline yet.
California’s SB 947 bars employers from relying solely on AI to discipline or fire workers. Public employers are covered. The hard part is a phrase the law never defines.
California’s SB 1246 spells out what robotaxi companies owe police, firefighters, and dispatchers when a driverless vehicle fails. Most of the rules apply by July 2028.
Tennessee requires approval for every AI tool state workers use. Now it is building monitoring to find and switch off the ones nobody approved.