Real question - have the big LLM purveyors collaborating on cross-validating their outputs? I do this manually all the time. I give one task to Grok or Gemini and then have ChatGPT or Claude go through and validate. This removes 99% of hallucination, and most citation gaffs.