@SpireSec

Tech Risk, Cybersecurity, and Digital Security.

Joined March 2011
The math doesn't work for researchers. Here is the most glaring example I've found where disclosure increases risk. (Btw, the collision arg holds no weight either.)
4
1
1
14
Replying to @OrinKerr
It seems clear from a legal point of view. It's also clear from a cybersecurity point of view. We want out-of-control AIs hacking the Internet as long as they aren't doing any damage. The more this happens, the more systems get secure, and the less threat from a superintelligent AI doing real damage.
6
6
33
12,176
My new paper on "The Value of a Statistical Life Year" finds that government agencies sometimes understate the VSLY. The VSLY varies with age, the discount rate, and how it is estimated. Climate change impacts are undervalued. papers.ssrn.com/sol3/papers.…
1
5
661
Here is the pretty fake FG ⚡️ #d3fb 🎥 Carnegie Mellon
Blue Streaks on the board early with fake field goal TD run by Shane Lindstrom! 📺 centennialconference.tv/?S=t…
18
105
37,901
An engaging rehash of the OpenAI/Hugging Face incident. If you got giddy reading that agents got giddy, keep in mind there was only one organic mass releasing dopamine, endorphins, etc. The other entities were simply providing algorithmic outputs based on inputs.
Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes. This culminated in the third one taking over part of OpenAI itself. All this happened while humans remained more-or-less in the dark about the scope of the conspiracy. I’ve spent the last three days reading through these reports and trying to understand exactly what happened. Here is my attempt to tell the whole story in plain English: dwarkesh.com/p/openai-huggin…
71
When are we going to find out that @pangram has special access to read all frontier llm responses? 😬
Genuine question. Why are AI-written law review articles objectionable? Say the author inputs the idea, research, detailed outline, etc., but AI drafts it. Now say the author drafts the same article the old-fashioned way. Why should we prefer (b) over (a)?
103
if you identify something as slop, then upload it to @pangram and it turns out not to be AI generated, is it still slop? does it get a pass because it's human slop? Isn't all slop derived from humans? I don't get this phenomenon. Thankfully, it will all be moot soon.
2
76
Marginal Revolution / Econlog / Econtalk; Against the Gods by Peter Bernstein; Kahneman / Tversky / Thaler; Structure of Magic I & II by Bandler / Grinder; Science of Influence by Cialdini; Crossing the Chasm / Inside the Tornado by Geoffrey Moore; Moneyball, etc by Michael Lewis
My son just posed a lovely question: What piece or pieces of media -- movie, song, book, TV show, article, poem, etc -- has really changed how you think? (How you think about life, about the world, or just how you think.) Would love to hear your answer.
61
This would make for a great @bryan_caplan bet...
I broke my own rule to never post about AI detection as it is fraught in many ways. The problem is that if you use AI a lot, you know AI writing on sight, which makes the difficulty of objectively proving that AI use to others very frustrating
80
New data from @OpenAI Economic Research: at least 4 million people used ChatGPT in March 2026 to help plan, start, run, or grow a business – a conservative lower-bound estimate. It’s an early sign of how AI can help more people access capabilities and expertise that were previously expensive, specialized, or out of reach. This is not just a startup story. Entrepreneurship in ChatGPT has strong activity across retail, professional services, trades, hospitality, health, and more.
9
27
5
120
7,412
A great episode.
NEW ODD LOTS: @tracyalloway and I talk to @alexolegimas about what economists might be getting wrong about AI and white collar disruption, and how to most clearly understand what jobs are at risk podcasts.apple.com/us/podcas…
46
NEW ODD LOTS: @tracyalloway and I talk to @alexolegimas about what economists might be getting wrong about AI and white collar disruption, and how to most clearly understand what jobs are at risk podcasts.apple.com/us/podcas…
10
36
19
249
228,667
Time to rethink defense. When patching is too slow, and assume for a moment it will be. Then what?
57
What would/will/has AGI written* about the identity of Satoshi? Who will believe it? [*what is the proper verb here? Disclosed, said, determined, concluded...]
27
Project Glasswing: fixing all the vulns... and patching all the instances... catching all the snowflakes before they can form an avalanche...Really? Seems doubtful to me. Now is the time to double down on runtime protection of all forms.
39
This is the last thing I'm going to say about Marcus Freeman and what happened with this "coach" from New Prairie. After this it's gone in my view. I'm disgusted and disappointed. I'm disappointed with the initial reporting and disgusted with the situation as a whole. The reporting was disappointing. I know the writer and I'm fond of him as a person, but he should have had editors and people above him that said "This isn't enough, we need more" or "Wait until the report is filed." Several people on the beat knew this story was out there, but we all made the proper decision to let the legal process properly play itself out. It's not only proper, it also protects us from looking bad when the true story, and the whole story, comes out. I'm disgusted by the "leadership" at New Prairie, who employs a man with such a tainted background, and for whom controversy is not new. If I knew anyone with any kind of authority at New Prairie I would be demanding he be fired immediately, and anyone that knows me knows I do NOT say things like that lightly. This should have already been done. Not only was this a false police report, this was targeted, and it's clear as day. Just read the report. If you know anything about this guy's background - which I do - you'd be appalled that someone would employ him and allow him to be around young people. It's appalling and New Prairie should be ashamed of itself as an institution. It's a shame that a young man like Vinny Freeman, who has a passion and talent for a sport gets a target placed on his back because of that ability, and because of who his father is. This isn't even about Marcus Freeman, who can take care of himself. But when you go after a man's children, you are true scum. The good thing is the justice system worked for Marcus Freeman and his family. They can put this behind them and know that this incident reflects on the "man" who went at him and doesn't reflect on him or his family. It's sad that a coach at a place like Notre Dame, other big time programs or NFL coaches, can't just go be a dad without dealing with this mess. That's it from me, time to move on and focus on the 2026 season, wish Vinny Freeman all the best the rest of the way and hope that he's able to enjoy his final season of high school, and that his parents get to enjoy being supportive parents for the rest of the season.
60
120
8
1,547
189,432
Not just straddling the fence, ND was in the CFP Top 10 every single week since November 4th with the same 2 losses it’s had since Week 2 (incl the H2H loss at Miami). Since then it went 4-0 and dropped out today after being idle, but Miami also went 4-0 and climbed 8 spots!
2
3
25
1,249
I don’t blame Notre Dame one bit for deciding not to play in a Bowl Game when they should have been playing for a chance to win the National Championship in the College Football Playoff.
3,061
903
209
11,529
524,328