Thursday 15 October 2026 5:03 am  |  Updated:  Wednesday 14 October 2026 5:53 pm

The worst thing we could do with AI right now? Slow down

By: Anna Moloney

Deputy Comment and Features Editor

Share

Facebook Share on Facebook X Share on Twitter LinkedIn Share on LinkedIn WhatsApp Share on WhatsApp Email Share on Email

AI doomsday headlines are naturally sparking fear, but the worst thing we could possibly do right now is slow down, writes David Grainger

The claim that we face a 10 per cent risk of AI destroying humanity within a decade has caused quite a response. Even the CEOs of Anthropic and rival OpenAI have joined calls for a slowdown in AI research while our understanding of these AI-powered agents and the damage they could cause catches up.

For many, this seems prudent. Maybe even essential. But it may be the wrong response entirely. Slowing down might even increase the risk to humanity that AI poses.

Why would AI agents destroy us? They don’t have motives like we do. The most benign reason is misinterpretation of an instruction – the cause of the unexpected hacking at Hugging Face.

A much more likely reason though is someone could build an agent with the express purpose of causing destruction. It is easy to demonise the technology, but whatever harm it can do, humans are worse.

........

© City A.M.