By now, every lawyer knows the artificial intelligence hallucination story: fake cases, sanctions orders, a public database tracking more than 2,000 court decisions and counting.

The profession has responded the way it should have — with verification duties, standing orders and cite-checking discipline — and hallucinated cases are now the primary AI error every lawyer knows to look for.

Less attention has been paid to the problem of AI sycophancy. This is the tendency of AI tools to tell users what they want to hear: to affirm their beliefs, adopt their framing and endorse their conclusions, even when accuracy requires disagreement.

Sycophancy is especially problematic in legal contexts because it reinforces confirmation bias, and it cannot simply be prompted away. Fortunately, there are ways lawyers can reduce it.

Read the full article written by Michael Ott and published in Law360. Login may be required.