Tech-backed group notes continued threat to open source in NIST’s revised guide to mitigate AI misuse

The National Institute of Standards and Technology’s AI Safety Institute has made changes to its guidance on abating the risk of AI foundation models being used for malicious purposes, but a notable tech industry coalition says they do not sufficiently address concerns the document could unjustifiably limit open-source developers.

“While we are supportive of several of NIST’s revisions, including the shift to evaluating models based on marginal misuse risk, the second public draft of AI 800-1 is a step backward...