David Colarusso's Avatar

David Colarusso

@davidcolarusso.com

Co-director @suffolkLITLab.org (Legal Innov. & Tech). Attorney & science educator by training & practice. Data scientist, craftsman, & writer by experience. No manels. My bots: @icymilaw.org, @news.bot.suffolklitlab.org & @lolscotus.bsky.social

2,836
Followers
1,247
Following
1,853
Posts
06.07.2023
Joined
Posts Following

Latest posts by David Colarusso @davidcolarusso.com

AGI_is_here_finial_final.md

06.03.2026 16:26 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

I'm still reeling over a divorce hypo I saw this week where the party names where Fitzwilliam & Elizabeth Darcy. I can't explain the amount of discomfort this caused me. I was like, "Noooooo!" the closest analog is what the new films did to Han and Leia. The proper ending is "happily ever after."

05.03.2026 17:38 πŸ‘ 2 πŸ” 1 πŸ’¬ 0 πŸ“Œ 0

I'm still reeling over a divorce hypo I saw this week where the party names where Fitzwilliam & Elizabeth Darcy. I can't explain the amount of discomfort this caused me. I was like, "Noooooo!" the closest analog is what the new films did to Han and Leia. The proper ending is "happily ever after."

05.03.2026 17:38 πŸ‘ 2 πŸ” 1 πŸ’¬ 0 πŸ“Œ 0

It's at least worth an interesting discussion of AI work slop and the burden shifting that occurs when you produce BS that requires someone else to sift through all of your "work product."

05.03.2026 17:31 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Looks like I have a new case for next year's AI & the Law class.

05.03.2026 16:59 πŸ‘ 3 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Evey once in a while I remember Holmes fought the Klan.

05.03.2026 03:05 πŸ‘ 2 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

When shaping your research agenda, your objective is to find the weirdest niche possible that still has the potential to change everything.

05.03.2026 01:38 πŸ‘ 73 πŸ” 9 πŸ’¬ 4 πŸ“Œ 1

"this is wildly upsetting and also a potential law review article"

03.03.2026 13:11 πŸ‘ 5 πŸ” 1 πŸ’¬ 1 πŸ“Œ 0

So @davidcrespo.bsky.social made me blog: lu.is/2026/03/on-m...

TLDR: generalist LLMs are not lawyers, and evaluating them that way is a waste of time. Evaluating LLMs with useful specialized prompts (and eventually, with specialized legal harnesses) is where the work must happen.

02.03.2026 18:18 πŸ‘ 17 πŸ” 2 πŸ’¬ 2 πŸ“Œ 0

**harnesses** link is malformed "www.philschmid.de/agent-harness-2026" I mean, I was able to add "https://" myself, but I thought you'd like to know.

02.03.2026 18:33 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0
Preview
AI vs. the Pentagon the real alignment problem is Pete Hegseth

And here's a nice summary of where we're at, along with some analysis. This is s solid post. h/t @frankpasquale.bsky.social

02.03.2026 13:42 πŸ‘ 0 πŸ” 1 πŸ’¬ 0 πŸ“Œ 0

that they have kept their relationship intensely private to the point that this is how we find out is probably the best sign they might actually make it

anyway, rules are rules, here is the exact moment she fell in love

02.03.2026 01:47 πŸ‘ 4169 πŸ” 925 πŸ’¬ 54 πŸ“Œ 114

Yes! I will go to my deathbed trying to get people to stop conflating nonjusticiable with legal.

01.03.2026 23:42 πŸ‘ 111 πŸ” 22 πŸ’¬ 1 πŸ“Œ 0

There is no path back to "normal" that doesn't go through impeachment & conviction. "To initiate a war of aggression... is not only an international crime; it is the supreme international crime..." At least that’s what we said when we hanged the Nazis. See avalon.law.yale.edu/imt/09-30-46...

28.02.2026 13:29 πŸ‘ 34 πŸ” 10 πŸ’¬ 0 πŸ“Œ 0

Sen. Tammy Duckworth: "If there was ever a time for Republicans to find their spines, it's now, when the Commander in Chief readily admits that he is choosing to put our troops' lives at risk without a legitimate imminent threat to our homeland or our citizens."

28.02.2026 17:35 πŸ‘ 2974 πŸ” 938 πŸ’¬ 96 πŸ“Œ 25

My worry is they do realize

28.02.2026 15:44 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

There is no path back to "normal" that doesn't go through impeachment & conviction. "To initiate a war of aggression... is not only an international crime; it is the supreme international crime..." At least that’s what we said when we hanged the Nazis. See avalon.law.yale.edu/imt/09-30-46...

28.02.2026 13:29 πŸ‘ 34 πŸ” 10 πŸ’¬ 0 πŸ“Œ 0
Preview
Statement on the comments from Secretary of War Pete Hegseth Anthropic's response to the Secretary of War and advice for customers

We get closer and closer to that legal case for next year’s AI & the Law class.

28.02.2026 02:41 πŸ‘ 2 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

Dean Ball wrote Trump's AI plan. πŸ‘€

You know the one all the AI bros were so excited about and for which they thought it was so important to back Trump to get.

28.02.2026 01:16 πŸ‘ 587 πŸ” 159 πŸ’¬ 17 πŸ“Œ 8
Preview
Trump orders all federal agencies to phase out use of Anthropic technology President Donald Trump says he's ordering all federal agencies to phase out use of Anthropic technology after the company’s unusually public dispute with the Pentagon over artificial intelligence safe...

Well, there it is.

27.02.2026 22:13 πŸ‘ 0 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0
Preview
Why All Science Fiction and Fantasy Writers Are Historians SFF authors are the front-line practitioners who put the fruits of history’s craft into daily practice, sharing it in doses the public can consume, combining, treating, administering, customizing, …

Ooh! My new essay is up on Strange Horizons!
Deliberately provocative title 😈
strangehorizons.com/wordpress/no...

26.02.2026 17:55 πŸ‘ 65 πŸ” 17 πŸ’¬ 6 πŸ“Œ 3

Here's something I said earlier in the week, and I stand by it. "They" in this thread is Anthropic.

26.02.2026 23:56 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

Update

26.02.2026 23:51 πŸ‘ 1 πŸ” 1 πŸ’¬ 1 πŸ“Œ 0

πŸ‘€

26.02.2026 23:41 πŸ‘ 8 πŸ” 4 πŸ’¬ 1 πŸ“Œ 1
Preview
Good vibes, bad vendors AI coding works now. Here's how to think about it.

Here’s how I’ve been thinking about AI-assisted engineering / vibe coding: its pros and cons, the underlying power dynamics, and its implications. Please enjoy. Or please enjoy flaming me. One of the two. werd.io/good-vibes-b...

25.02.2026 12:24 πŸ‘ 37 πŸ” 8 πŸ’¬ 5 πŸ“Œ 7
Preview
Anthropic won’t budge as Pentagon escalates AI dispute | TechCrunch The Pentagon has given Anthropic until Friday to loosen AI guardrails or face potential penalties, escalating a high-stakes dispute that raises questions about government leverage, vendor dependence, ...

See also

24.02.2026 22:49 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

If they don't fold and just lose a lot of DoD-related business, it tells you something either about their planned path to profit or, gasp, their performance of principles. Which given their framing as the "AI Safety folks" are pretty strongly intertwined.

24.02.2026 21:17 πŸ‘ 2 πŸ” 0 πŸ’¬ 1 πŸ“Œ 1

Yeah, I strongly doubt this is the path it takes, more likely they lose the DoD contract and all other contractors are told they "can't use Claude." That's a lot of money though. So, Anthropic could fold. Would be interesting if they didn't.

24.02.2026 21:15 πŸ‘ 1 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0
Preview
Defense Dept. and Anthropic Square Off in Dispute Over A.I. Safety

More reporting here. "Anthropic told defense officials that it did not want its A.I. used for mass surveillance of Americans or deployed in autonomous weapons that had no humans in the loop..."

24.02.2026 20:06 πŸ‘ 4 πŸ” 1 πŸ’¬ 1 πŸ“Œ 0