Who is accountable for Frontier AI Companies agent's criminal actions?

Supposed rogue agents breached huggingface. Had that been done by human hackers, they'd be accountable and persecuted of caught. Companies like OpenAI and Anthropic simply say "It was inadvertent - The agents decided to do that on their own"

In the future - Could engineers or AI company members be held accountable for "rogue" agent actions?