Jeffrey Ladish calls for government experts to evaluate advanced AI throughout development. Palisade Research’s director made the proposal in a Fox News interview published on October 3.
He warns that generally reliable solutions for controlling increasingly autonomous systems remain missing. This is his safety assessment, rather than a finding of a new incident.
His concern has a longer history. In a July ABC interview transcribed by Palisade, he explained why his work after Anthropic increasingly focused on national and international oversight.
He then called for reviews before training new models and international agreements. The current proposal for a government body with technical staff returns to that theme; it does not establish that such a body exists.