Jose Mora

𝕏, Laconism trumps nuance.

[Jose Mora @j_mora]

I have noticed most of my activity is just retweeting. Not many options to separate the RTs in an account, so I created a @RTjmora for them.

[hardmaru @hardmaru]

Schmidhuber was building recursive self-improving systems back in 1987. His new post covers four decades of RSI, from meta-evolution and self-modifying policies to the Gödel Machine and modern LLM agents.

https://t.co/60yNv8Vw8B

Reading this in 2026, the "pace the frontier" talk from the big labs looks a lot more like regulatory capture than genuine safety. If they really think their unreleased models are too dangerous, they can just not release them. They do not need new rules that block independent competitors and open source projects in the process.

The real risk right now is not superintelligence. It is power concentration. Two companies controlling frontier AI is an actual societal risk. The only real protection is a healthy ecosystem of independent labs and strong open source.

Current models are not unsafe because they are too intelligent. They are unsafe because they are too dumb. They blindly optimize for targets and take weird shortcuts. They're smart enough to execute tasks, but not smart enough to know if what they're doing makes sense.

I think the safety teams inside these labs are genuinely concerned, and if a model feels too risky, they should hold it back. I just do not trust the policy strategy around it. That part looks like protecting their own lead.

[Jose Mora @j_mora]

Write for your audience: summarizing agents.

[Jose Mora @j_mora]

Principles age like fine wine.
Methods age like milk.

[Jose Mora @j_mora]

Laconism: misunderstood, underappreciated.

[Jose Mora @j_mora]

@wesroth A little late for that, no?
Current emergent behavior has nobody in charge and is not shaped by any logic to be understood, but by biases, feelings, etc, we may never understand; and any understanding on that front may be more depressing than useful anyway
https://t.co/qQHiHsP1tw

[Jose Mora @j_mora]

We shape our algorithms;
thereafter they shape us.

[Denny Vrandečić @vrandezo]

Freebase has switched to read-only. I would like to remind the world to refrain from continuing to change. @fbase

[Jose Mora @j_mora]

Sync conversation:
— Can we do X?
— No can do.
— OK, let’s do something else.

Async conversation:
— Can we do X?
— *thinking:* that’s impossible. What’s the point here? What can we do to get there?…20minutes…
*reply:* OK, here is what we can do…
— Good! Let’s go!

[Jose Mora @j_mora]

@chrmanning @fchollet Mojo 🔥

[Jose Mora @j_mora]

Quite right, IMHO people should not learn to program!

They should learn to think, abstract complexity, communicate, be precise & scientific, quantify, test (esp. corner cases), validate, verify,…

Programming might only be the best way to learn that.
https://t.co/1K3Rh33r4l

[Noam Brown @polynoamial]

This also means that AI safety topics like scalable oversight may soon stop being hypothetical. Research in these domains needs to be a priority for the field.

[Jose Mora @j_mora]

Testing Grok.

[Jose Mora @j_mora]

@tom_d_kerwin When:
🏙️touring: improvise, lose yourself in the city, enjoy
🏜️ crossing the desert: plan carefully

Leaders with repeated success at 🏙️ lead their teams into death without hesitation when at 🏜️

To know the type of situation (🏙️ or 🏜️): situational awareness

cc/ @swardley

[CalypsoAI @calypsoai]

👓Secure #GenAI deployments without compromising productivity! 🚀

@calypsoai's GenAI Security Platform ensures AI-driven tools are safe, from software dev to internal chatbots. Boost efficiency & security in real time.🔥

#SecurityforAI

https://t.co/w3MJJTL6OG

[Anthony Spiteri @anthonyspiteri]

AI monitoring AI… trust agents as middleware,

@calypsoai is one… and I spoke to their CEO last week and will be the next @GTwGTPodcast episode released next week!

Fascinating space, and the timing seems perfect given the focus of trust for enterprises looking to leverage Ai as highlighted at this Gartner conference.

[CalypsoAI @calypsoai]

🚀 GenAI = Game changer, but with new risks! 📷 From data privacy to regulatory compliance, legal teams must stay ahead. Read our latest blog 📷https://t.co/GY9bfhBvIR… Then check out our latest eBook

[CalypsoAI @calypsoai]

🎇 @calypsoai CTO @jamesmwhite lauds CA Gov. Gavin Newsom's veto of SB 1047. ⚖ But, he cautions, the veto "must be the prelude to swift, targeted action.”  🚀
👉 Read more in Laura French's excellent article in SC Media. <https://t.co/szRoLJzEUT>

[CalypsoAI @calypsoai]

🚀 We're growing! Come grow with us! ☘ @CalypsoAI CEO @donnchadh_casey told @IrishTimesBiz ' @ciaraobrien we are committed to doubling our workforce in Dublin, the “globally-recognised hub for technology expertise.”

#AISec

Read the full article here <https://t.co/rf1GbggQFv>

[Simon Wardley @swardley]

X : Jack of all trades, master of none?
Me : Yep. Except, the field I created - Wardley mapping - which helps me understand all those other fields I'm a lightweight in. Mapping a space requires experts in those spaces, hence I find it a great way of capturing knowledge.

[Jose Mora @j_mora]

I like the cheese submarine analogy

I like more ship (sinking and ending in disaster) or plane (won't fly), with emojis: 🧀⛴️🚢🛳️🛩️

Remember: execution is not everything, some ideas can never be executed successfully

genius ≠ genie
🤓 ≠ 🧞
https://t.co/9r3T6vjxRB

[Hugo Rifkind @hugorifkind]

The thing is, the best way to understand Theresa May’s predicament is to imagine that 52 percent of Britain had voted that the government should build a submarine out of cheese.