Post hoc detection analyzes content after production to judge whether an artificial-intelligence system created it. For large language model (LLM) text, reliable detection remains unsolved: increasingly fluent, humanlike outputs narrow the gap between generated and human writing. section 10.23
Detection can support accountability in education, publishing, and public communication, but it should not be treated as conclusive evidence. An OpenAI classifier showed low accuracy and was discontinued in July 2023. Image, audio, and video generation may remain detectable in some cases, although those gaps are also closing. section 10.23
Watermarking offers a different route by embedding proof of machine generation into output. It would require broad adoption and public verification, yet users could alter text to evade checks, attackers could imitate watermark patterns, and quality could decline. section 10.23
