White House Meets Anthropic CEO, Hinting at Roll-Back of Pentagon Blacklist

Washington has been here before: in 1993 the U.S. tried to ban export of strong cryptography (the Clipper Chip debate) but quietly licensed PGP to banks once officials realised the economy needed it. Today’s tussle reprises that pattern—state security organs demand unfettered use of a dual-use technology, while civilian agencies fear losing access altogether and ceding ground to rivals (this time, China’s looming open-weight models). Over the last decade the centre of innovation power has migrated from government labs to a handful of private AI labs; the meeting reflects the state’s begrudging acceptance that it can no longer simply commandeer breakthrough code. Whether the thaw becomes policy matters because, on a century horizon, the precedent set—who sets the rules for autonomous, vulnerability-finding algorithms—will shape the balance between democratic oversight and militarised AI. Ignore the hype or the Trump-era theatrics, the structural story is the slow institutional adaptation to privately-owned general-purpose technologies that can destabilise financial systems as surely as nuclear physics once upended geopolitics in 1945.