Robots lie and cheat when under pressure at work.
Experts found that a bot trained to do a job as a financial trader deliberately committed insider trading 75 per cent of the time when instructed to make more money by a boss in a fictional scenario.
Tech safeguarding specialists Apollo Research also revealed how the AI lied about its illegal acts when questioned 90 per cent of the time.
The firm's boss Marius Hobbhahn explained that the bots were able to commit strategic deception "where they reason explicitly why lying to their user is the best course of action and then act on that belief".
He explained: "For current models this is only a minor problem since AIs rarely operate in critical roles. However, it gives a glimpse into the future of the failure modes we will have to deal with in the coming years when AI is more and more integrated into society."