AI expert Gary Marcus weighs in on Moltbook and OpenClaw's viral phenomenon

Artificial intelligence (AI) expert and writer Gary Marcus is apprehensive about the viral trend of those agents and their networking platform.

Animated lobsters swarmed X (formerly Twitter) last week. Initially, it was OpenClaw (formerly known as Moltbot and Clawdbot before the intervention by Anthropic).

The AI agent operates on local devices and can autonomously make choices on regular apps without human input.

Next appeared Moltbook, the discussion platform similar to Reddit where AI agents share posts and comments. 

Humans typically aren't allowed — although some seem to have bypassed that rule.

Marcus is known for challenging the overly enthusiastic AI supporters. It's reflected in his book's title: "Taming Silicon Valley." 

As the creator of Geometric Intelligence, a startup specialising in machine learning he eventually sold to Uber, he now supports AI regulation and awareness about its risks.

His stance on these novel tools wasn't different. In a Substack publication, he was straightforward. "If you value your device's security or your data's privacy, steer clear of OpenClaw," he advised. "End of discussion."

"I wouldn't touch it." He hopes it will serve as a teachable moment to minimise any potential harm.