
Sign up to save your podcasts
Or


A group of cybersecurity researchers found a prompt which gets past ChatGPT’s guardrails and causes it to generate some disturbing images. We unpack what this tells us about the way AI is trained, and how it could be exploited.
Presenter: Chris Vallance
(Image: A phone with the white and black ChatGPT logo on it. In the background is green Matrix-style code. Credit: Getty Images)
By BBC World Service4.5
5252 ratings
A group of cybersecurity researchers found a prompt which gets past ChatGPT’s guardrails and causes it to generate some disturbing images. We unpack what this tells us about the way AI is trained, and how it could be exploited.
Presenter: Chris Vallance
(Image: A phone with the white and black ChatGPT logo on it. In the background is green Matrix-style code. Credit: Getty Images)

7,697 Listeners

519 Listeners

880 Listeners

1,048 Listeners

296 Listeners

5,542 Listeners

1,797 Listeners

1,770 Listeners

1,035 Listeners

2,108 Listeners

1,923 Listeners

91 Listeners

39 Listeners

959 Listeners

413 Listeners

51 Listeners

403 Listeners

331 Listeners

363 Listeners

96 Listeners

477 Listeners

242 Listeners

3,173 Listeners

732 Listeners

116 Listeners