I worked at LexisNexis Risk for years on the engineering side of the credit scoring, fraud detection, identity verification, etc. That was almost ten years ago,

I worked at LexisNexis Risk for years on the engineering side of the credit scoring, fraud detection, identity verification, etc. That was almost ten years ago, but I suspect the core value prop that LN offers here remains unchanged: 1. Buying like datasets and combining them into a single, national dataset, and 2. Linking those datasets together The former is a challenge for anyone to do because you have data like property records that come from county assessors. Vehicle data - which is what seems to be relevant to this article - may be at the state level, which is easier than county, but it's still dealing with 50 states instead of one national registry. The latter is important because connecting "JON SMITH" in one dataset with "JOHN SMITH" in another is a nontrivial problem that LN has addressed quite well by virtue of many different datasets. There are (or at least were; with insurance data, this may be different) broadly two categories of data: FCRA-encumbered and non-FCRA data. FCRA data is more heavily restricted, and you're entitled to correct data that are wrong. As I recall, there are also legal limitations on who can use these data and for what reason. So my assumption is that not only is ICE buying this, but they also have to assert that they are legally allowed to do so. I'm not saying I'm cool with this contract, but in lieu of LN choosing to not to business with ICE, there's also the option that legislators could make it illegal for them to do this.