‘The AI did it’ just isn’t an accountability technique
Consultants condemn Railway’s “rogue AI” clarification for the incident.
Amir Boldo, co-founder and chief product and expertise officer of Above Safety, cautioned in opposition to describing brokers as attackers, as they will execute attacker-like sequences with none malicious intent.
Execution traces can set up what an agent did, however they can’t independently show intent, he stated, including that investigators want proof spanning directions, identities, permissions, inputs and power calls. “You possibly can’t put an agent on trial. You possibly can maintain every occasion accountable for its half,” Boldo stated.
