Ask HN: Does ChatGPT 5.2 with web search still hallucinate facts?
We all know that AI's have a hallucination problem. We have to check sources. Lately, I've noticed that ChatGPT 5.2 with web search cites its sources very well, and I've stopped noticing hallucinations. Are the hallucinations just more subtle now? Does it still hallucinate and if so, can you give recent examples?