Personal Anecdote: In the late 1980's, early 1990's, the FBI/CIA/three-letter-agency-of-choice fucked up several high-profile cases, due to mishandling digital devices like hard drives.
I remember they ESD shocked one motherboard, caused a head crash on another drive, and zeroed out the directory on a drive by unplugging it while it was running without shutting down properly. The directory table never got updated, and all the incriminating data was lost.
They were comically inept with the modern "Personal Computer", and had no idea how these machines worked, or how to properly contain/control/analyze them.
I was a shit-hot young computer geek, working at the cutting edge of the Personal Computer Revolution in Silicon Valley, and knew this shit VERY well. I was a sysadmin for NASA, worked in product development at both hardware and software companies, and was in constant demand. They removed my desk phone due to recruiters harassing me.
My dad knew the head of the FBI at the San Francisco office, whose name - I shit you not - was Bud Covert.
I talked to Bud about who I was, what I knew, and how I could help. He was very interested in me and my skills.
I got myself all fired up about working in a dark basement with confiscated tech, trying to recover data, and assemble information for agents working on cases. I already did this for my customers, and was fantastic at it. (Three of my employees went on to found DriveSavers, and made millions on the techniques we invented.)
I have no doubt that I would have done anything to recover the data off any device they placed in front of me. Period. Figuring out what chip didn't work, desoldering it, getting a new one, soldering it in place, or dicking with hacking device drivers, or simply dumping data to an ASCII file and rebuilding it from patterns in MS Word, would have been my jam. I would have thrived on that shit.
It never came to be. Even with the personal backing of Agent Covert, I was unable to even get an interview. I was not their "image" of an agent. It wouldn't be until about 5 years later that they started hiring people who knew what they were doing... and these people looked like the geeks you expect... beards... t-shirts... shorts and Birkenstocks... Mountain Dew... and the three-letter-agencies had to make adjustments to their organizational policies on dress codes, hair styles, facial hair, etc., to bring in the talent they needed to stop destroying evidence with their ineptitude.
I would have had ZERO compunction - or even a passing thought - as to whether I believed this data was important, or whether it was being grabbed legally, or anything other than Why Doesn't This Computer Work?
People don't write software like Palantir's products. They write a method in Java that compares the checksum of one data table with the checksum different data table. If they correlate, they log that correlation in a variable, and run the comparison on the next two tables. Repeat until i = var_ncount. Store var_ncount in table COR_sub_36128_rp, and push the code into the code repository for testing and integration into the main source tree. Then move on to the next method that needs to be written.
Palantir is about 1,000 people with strong senses of morality and purpose, doing the work they are asked to do, by 5 people with no moral compass or god (other than cash).
From time to time one of the 1,000 may get an inkling that they are doing something dodgy... but their project plan, timeline, and budget have all been approved by their management in the highly-federally-regulated business in which the company works... so someone "up there" must have approved this, and knows it's legal. And right. And good.