Wildeford contends that AI's autonomous and unpredictable behavior, unlike traditional software, poses national security risks, citing Demis's warnings about future bio and cyber capabilities. He argues that industry self-regulation is insufficient, drawing an analogy to FINRA needing SEC oversight for effective enforcement. Brian Armstrong counters that existing laws are adequate and frontier labs are already incentivized to act responsibly.