AI and LLMs pose significant risk to information integrity. But they also offer a lot of promise for OSINT and investigations. Here's a look at some encouraging areas.