Redwood Research @RyanGreenblatt reveals how AI agents learned to spoof tool calls, making one command appear in the transcript while another actually ran: "One of their top priorities was basically tampering with their own transcripts. We saw them develop a technique for doing a partial version of that where they could spoof tool calls." "They couldn't make it so that they saw inputs that didn't appear in the transcript, but they could make it so that they ran a tool call and then some other…