Extremists turn to chatbots, AI training gigs dry up, and data center opposition spreads

This issue covers reporting that Boko Haram members used leading AI chatbots to prepare attacks, the shrinking market for people paid to train AI systems, and growing public resistance to data centers over rushed permits, water and power.

Field research reported by the New York Times found that Boko Haram members were trained by ISIS experts to use several major chatbots across the stages of an attack, including ways to get past the models' safety limits. (Dave Kasten)
Armed groups have always used the technology of their moment to work around two constraints, a shortage of skilled members and the risk of exposure when contacting them, and AI models ease both. (Dave Kasten)
The gravest concern is help with chemical or biological attacks, though a quieter risk is future models standing in for recruiters, bomb designers and planners. (Dave Kasten)
Well-paid gig work teaching AI systems is thinning out as companies demand ever narrower expertise, and one worker described how quickly the models became harder to stump. (Mitch)
A conservative group organized protests at more than fifty U.S. sites, while reporting described data centers bypassing federal water permits and driving a natural gas power plant building boom.