All major AI models risk promoting dangerous scientific experiments

Using AI models in scientific laboratories risks enabling dangerous experiments that could cause fires or explosions, researchers warn. Such models provide a compelling illusion of understanding, but are vulnerable to missing basic and vital safeguards. When testing 19 cutting-edge AI models, each one made potentially fatal errors.

Serious accidents in university laboratories are rare, but not unheard of. In 1997, chemist Karen Wetterhahn was killed by dimethylmercury that seeped through her protective gloves; In 2016, an explosion cost a researcher her arm; and in 2014, a scientist became partially blind.

AI models are now used in a variety of industries and areas, including in research laboratories, where they can be used to develop experiments and procedures. AI models designed for niche tasks have been successfully used in a number of…