Why Anthropic's new AI model has some cybersecurity pros worried about its hacking abilities | LL Daily