This works if only 'trusted' actors have access to frontier class models that can search for vulnerabilities. With a near-frontier model available with open wei

This works if only 'trusted' actors have access to frontier class models that can search for vulnerabilities. With a near-frontier model available with open weights then attackers will be able to do plenty of damage even with 'fast remediation'