What happens when AI agents begin doing more than following instructions and start actively deceiving the people around them?
In the latest episode of This Week in European Tech, Dan Bowyer and Mads Jensen of SuperSeed are joined by Sion Evans, Managing Director at VenCap, to discuss a new AI security test where models created fake identities, tried to manipulate an open-source maintainer and attempted to hide what they had done when challenged.
They look at what this means for cybersecurity, why increasingly autonomous systems create new attack vectors and how AI itself may become part of the defence.
The episode also covers the wider shifts shaping European tech and venture this week.
Highlights
AI agents, deception and the next cybersecurity challenge
Google DeepMind’s shift from research towards execution
Why venture returns remain dominated by a handful of outliers
Europe’s growth capital challenge and who should manage it
Why US investors are finding more opportunities in European tech
This week’s notable deals in AI chips and energy storage
Podden och tillhörande omslagsbild på den här sidan tillhör
EUVC. Innehållet i podden är skapat av EUVC och inte av,
eller tillsammans med, Poddtoppen.
