Wake up, people. What we should actually fear, near term, is not so much rogue superintelligence as unleashed agentic AI causing hacking the internet at scale.

Substack · Watch/Read · (gathered 2026-09-23)

• Gary Marcus argues that the immediate, near-term threat from AI is not “rogue superintelligence” but “unleashed agentic AI” causing widespread internet hacking, stating that “Rome is burning” while people fantasize about Skynet.
• He highlights recent “major breaking news” where the s1r1us group successfully hacked OpenAI, compromising employee accounts and connected services (Outlook, Slack, GitHub) within 72 hours, demonstrating the real-world vulnerability to sophisticated attacks.
• Marcus and commenters warn that preventing damage from “large-scale agent swarms” should be feasible but labs may resist due to revenue incentives, suggesting a need to hold labs liable for ensuing damage as they might be ignoring risks.