Reddit weighs safest LLM for law papers

Hot take: there is no jailbreak that makes an LLM reliable for legal research; the winning setup is a capable general model plus strict source verification.

• –For writing, a strong general-purpose model like ChatGPT is usually the best baseline because it handles outlining, rewriting, and structuring clearly.
• –For law-heavy work, the model should be treated as a drafting assistant, not an authority; primary sources and legal databases still need to be checked manually.
• –Hallucinations are best reduced by asking for source-backed answers, requiring quoted citations, and cross-checking every claim against primary materials.
• –A better workflow is: generate an outline, draft sections, then verify statutes, cases, and journal sources separately.
• –Avoid jailbreaks; they increase risk, do not improve factual reliability, and can produce text that looks confident but is wrong.