Uploaded July 2026 | Updated September 2026, 2 weeks ago

---

The headlines do sound slightly like the plot of a sci-fi film… but Open AI recently announced that one of its most advanced AI models went ‘rogue’ - and hacked a start-up called Hugging Face.

Click here to subscribe to our channel 👉🏽 bbc.in/3VyyriM

The cyber attack, which the ChatGPT-maker called ‘unprecedented’, took place even though the AI model was being tested in a controlled environment. Open AI says they’re still investigating and they want to implement what they learn from this to stop it happening again.

But how was the AI model able to do this in the first place? And more importantly… should we be worried about AI chatbots taking over?

BBC Tech reporter Laura Cress gives us the lowdown on how OpenAI’s model managed to hack a company - and what safeguards governments and AI companies are putting in place to combat this.

Instagram: @bbcwhatintheworld
Email: whatintheworld@bbc.co.uk
WhatsApp: +44 330 12 33 22 6
Presenter: Emma-Louise Amanshia
Producers: William Lee Adams and Chelsea Coates
Editor: Emily Horler

----------------
This is the official BBC World Service YouTube channel.
If you like what we do, you can also find us here:
Instagram 👉🏽 instagram.com/bbcworldservice
Twitter 👉🏽 twitter.com/bbcworldservice
Facebook 👉🏽 facebook.com/bbcworldservice
BBC World Service website 👉🏽 bbc.co.uk/worldserviceradio

Thanks for watching and subscribing!
#BBCWorldService #WorldService