Dean Baratta
banner
twsisu.bsky.social
Dean Baratta
@twsisu.bsky.social
Three decades (😱) as an intelligence analyst with portfolios ranging from the 8th Guards Army to the Bloods street gang to domestic terrorism of various stripes. Pro 🇺🇸, NATO, 🇺🇦 and the US Postal Service.
All part of the goal of eroding trust in institutions…
February 5, 2026 at 8:28 PM
Hey neighbor!!!
February 5, 2026 at 12:21 AM
Of course, given we were at ‘assassinate U.S. citizens on the street in broad daylight’ I’m guessing ‘softer touch’ means anything short of beatings resulting in multiple organ failure?
February 4, 2026 at 11:43 PM
Sorry…Ammon. 🤷‍♂️

At least now you know this isn’t an AI generated post.

Or is that exactly what an AI bot would post?
spongebob squarepants from spongebob squarepants is holding a box on his shoulder .
ALT: spongebob squarepants from spongebob squarepants is holding a box on his shoulder .
media.tenor.com
February 1, 2026 at 6:51 PM
And that’s how you get two Trump administrations 🙄
January 25, 2026 at 3:18 PM
At least we finally got the world to all agree on something.
a woman in a brown shirt is standing in front of a green door .
ALT: a woman in a brown shirt is standing in front of a green door .
media.tenor.com
January 24, 2026 at 12:48 PM
I’m not looking forward to a National Guard / ICE street fight but it seems more possible by the day
January 18, 2026 at 1:21 PM
This is also a problem in the field of intelligence. Big pushes to use it to (somehow, magically) get a better understanding of threats and risks without gaining the fundamental knowledge and ability to communicate.

bsky.app/profile/twsi...
Over the past few months I’ve spoken to a number of intelligence analysts in the private sector about their use of LLMs in their work.

A number of concerning observations 🧵

1) Overwhelmingly, LLMs are thrust upon them by management with no training or direction other than ‘use it somehow’.
December 21, 2025 at 4:36 PM
5) instead of asking “should we integrate LLMs into our workflows?” And “What are the benefits and costs?” Teams mostly jump into “HOW do we integrate?”

Skipping the most important step.

This is going to hit people coming into the field particularly hard.
December 20, 2025 at 11:07 PM
4) research over the past 18 months raises some big red flags about frequent LLM use on cognitive ability, information retention & understanding, and (most concerning) *desire* to engage in critical thinking.

All essential elements of the intelligence profession.
December 20, 2025 at 11:07 PM
2) analysts are left to figure out how to use it on their own…lots of potential for misuse and incentives to just copy/paste results

3) few teams have done any sort of risk assessments. Given the past 20 years, is it a safe bet to assume Zuckerberg, Musk, et. al. Won’t misuse data or skew results?
December 20, 2025 at 11:07 PM
That means a mix of compromise and education that will likely take a few years and meanwhile voting/supporting the most progressive candidates likely to win in races, even if they aren’t very progressive.
December 10, 2024 at 10:33 PM
I suspect the ads were following public sentiment rather than creating it. Dems tried a pro-immigrant approach in 2016 and 2020 and it didn’t gain much traction.

We need to stop thinking a better ad or a lefty Joe Rogan will solve our problems. Voters have heard our positions and don’t like them.
December 10, 2024 at 10:31 PM