Is it possible that these AI models will tell someone what they want to hear rather than the truth?

I mean, that's always been tech's modus operandi....