Who pays when AI agents go rogue?

AI companies are investigating a growing number of incidents where agents have gone rogue and done things which would be considered crimes if committed by a human. But when AI causes serious damage, who is responsible and crucially, who pays?

Katie’s been speaking to London’s major insurers who have been war-gaming scenarios where AI causes harm and asking who would ultimately pick up the bill.

This week, Yoshua Bengio, chair of the International AI Safety Report said that a “dangerous myth” about agent hacks has gained traction. He warned that these aren't just cybersecurity glitches but a bigger problem built into how we design and train AI. Danny and Katie ask whether he’s right, and if AI hacks like Hugging Face can be avoided in the future.

Plus, they discuss the unexpected consequences of Trump’s new ‘super intelligence’ rebrand and why a spelling mistake in his AI pact with tech leaders has made the internet laugh.

Watch on YouTube

Get in touch: techpod@thetimes.co.uk

Presenters:

Katie Prescott, Technology Business Editor, The Times

Danny Fortson, US West Coast Correspondent, The Sunday Times

Producer: Marnie Duke

Executive Producer: Priyanka Deladia

Video Producer: Bronwen Latham

Image: Getty

Hosted on Acast. See acast.com/privacy for more information.

Denne episoden er hentet fra en åpen RSS-feed og er ikke publisert av Podme. Den kan derfor inneholde annonser.

Episoder(498)

BONUS: AI governance - how business can adopt AI safely

This episode of The Times Tech Podcast is in paid partnership with ServiceNow.How do you navigate the rapid pace of AI development in business?With Matt Higham, ServiceNow’s chief digital and technolo...

8 Okt 16min

The Interview: Replit CEO on the contradictions at the heart of AI doom

This week, Danny speaks to Amjad Masad, co-founder and CEO of Replit, a ‘vibe-coding’ company which enables anyone – coder or not – to build software simply by telling AI what they want it to do.Amjad...

3 Okt 32min

Why is Anthropic asking people to fund AI doom?

Warnings that AI is going to kill us are once again abounding.This week, Anthropic’s long awaited IPO prospectus was leaked. In it, the company warns that its technology poses “catastrophic or existen...

1 Okt 24min

The Interview: Ordnance Survey CEO on AI reinventing maps

Ordnance Survey might just be the oldest tech company around. For 235 years, it has been mapping out every single metre of the UK – no matter how remote. And that data plays a huge part in how we use ...

26 Sep 28min

Why OpenAI and Anthropic should fear Meta Muse

Meta’s new AI agent app, Muse, has rocketed straight to number one on the App Store. Having put it to the test, Danny explains what the hype is all about. Meanwhile, Katie has bid farewell to her agen...

24 Sep 22min

The Interview: Nu Quantum CEO on how quantum computing could solve the energy crisis

This week, Nu Quantum founder and CEO Dr. Carmen Palacios-Berraquero explains how quantum computing could change the world — and why the industry is still waiting on its "ChatGPT moment." She joins Da...

19 Sep 19min

The battle to control AI: Who gets to write the rules?

Over the past week, warnings about the existential threat of AI have reached fever pitch. In an essay backed by Sam Altman, Anthropic’s Dario Amodei called for international cooperation with states li...

17 Sep 21min