President Joe Biden is going to sign a memo on this Thursday outlining how US intelligence and national security agencies, including the Pentagon, would make use of and deploy guardrails around AI, The New York Times reported.

Orders the people to keep humans "in the loop" of AI tools that may be used as targeting weapons and bars allowing AI to determine who receives asylum, to track someone based on his ethnicity or religion, or to classify a person as a "known terrorist" without human review.

More than this, the memo requests that intelligence agencies begin protecting work on AI and AI chips from spying or theft by foreign adversaries. It also gives new powers to recently established AI Safety Institute to inspect AI tools before their release to ensure that they could not aid terrorist groups or unfriendly nations.

As The New York Times writes, though, it's not clear just how powerful the order will be in practice, as most of the deadlines he sets expire before Biden leaves office.