Data Poisoning Testing Techniques

I'm researching best practices for testing data in LLMs, especially for highly regulated industries like life sciences, financials, etc. If anyone has suggestions for recent articles or white papers on the techniques deployed in the recent months, would be appreciated.