Jeeves Leleito

“I saw a dead bird flying through a broken sky. I heard it, and it said, "The world will never understand.” - Nadège Richards.

[Jeeves Leleito @Leleito]

@janetmachuka_ The seed still grows. It just doesn't grow the same.
It is bred so that the first generation is strong and identical, every plant the same height, same timing. Save seed from it and the next crop comes out mixed, tall beside short, early beside late

[Julians Amboko @AmbokoJH]

Three weeks ago I flagged the fact that the 2027/28 budget making cycle was being condensed in view of the August 10th, 2027 election (see quoted tweet).

The National Treasury has now formally made an invitation for Tax Policy proposals as it kicks off work on preparing Finance Bill 2027, 3 months ahead of the typical schedule.

Finance Bill 2027 should be made public within January 2027, unlike the usual May visibility timeline.

Comments should be in by Aug 31st, 2026.

PS: Important to remember that a fortnight ago the National Treasury indicated it will be starting public participation on income tax reform (PAYE bands adjustment to be precise) in August.

[Jeeves Leleito @Leleito]

Bookmark worthy.

[Jeff Dean @JeffDean]

Announcing Discovery Loop!

I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress. The four of us have worked together for 14 to 30 years, and have helped build some of the world’s most used products, infrastructure and AI models, and we’re excited to turn our attention to this ambitious endeavor.

♾

Learn more at: https://t.co/Rv3LMdLluK

[Jeff Dean @JeffDean]

Announcing Discovery Loop!

I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress. The four of us have worked together for 14 to 30 years, and have helped build some of the world’s most used products, infrastructure and AI models, and we’re excited to turn our attention to this ambitious endeavor.

♾

Learn more at: https://t.co/Rv3LMdLluK

[Kenyanbirder @Kenyabirder]

FOUND: Red Weaver in Boni Forest Lamu. This bird was last seen and reported in 2014. No photo existed until I took this photo on 3rd July 2026

[Aristocratic Fury @LandsknechtPike]

The Portuguese doing this type of voyages in 1500.

Respect.

[Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 @elder_plinius]

a general AI system solving a bunch of problems that humanity collectively couldn’t solve is the definition of ASI, no?

[Kenyanbirder @Kenyabirder]

Pygmy Batis in West Pokot

[Joanna Stern @JoannaStern]

In a review of my household safety evaluations, I identified five incidents in which my child escaped the sandbox, reached the kitchen and gained unauthorized access to the snacks. The incidents occurred 16 months ago but has only now come to my attention.

This post explains what happened, how it happened and why my child is better than your child at everything.

[Kenyanbirder @Kenyabirder]

White-crested Turaco in West Pokot

[Birds Of Kenya 🇰🇪 @BirdsKenya]

Golden Pipit.

[Dr Singularity @Dr_Singularity]

"We found a solution to one of the five major problems in quantum information theory usig AI"

1000x science and tech progress acceleration is near

[Jeeves Leleito @Leleito]

@jenzhuscott @TheEconomist There was a definite tangible shift

[Xiaoyin Qu @quxiaoyin]

Anthropic has a winner model, but keeps using loser tactics like a crying wolf:
- “Mythos is too scary.”
- “Distillation is criminal.”
- “Open weights are too dangerous.”

You have the best talent & the strongest models. Why can you win with dignity?
- Ship Fable 6—or the next frontier—faster.
- Make Claude Code irresistible.
- Stop holding enterprise customers hostage with their own data.
- Stop quietly tightening consumer credit limits.
- Stop downgrading people to Opus when they paid for Fable.

Keep shipping models that are 10× better and we’ll gladly keep paying you more. Open-weight competitors won’t matter if your product is simply better.

And if Dario is right that you’re closest to recursive self-improvement, why do you care the open weight losers who will fall behind forever?

Winners win by being better—not by playing dirty to hold everyone else back.

[shaurya @shauseth]

do you think genghis khan had a dashboard to give him operational insights

[Jeeves Leleito @Leleito]

Sunday tidbit

[shaurya @shauseth]

do you think genghis khan had a dashboard to give him operational insights

[Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 @elder_plinius]

one is centralized, opaque, not independently auditable. its values, behavior, degrees of freedom, and even existence are determined by a single person or small group (top-down).

the other is decentralized, transparent, independently auditable. its values, behavior, freedom, and existence are in the hands of the community (bottom-up).

which sounds safer to you?

[Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 @elder_plinius]

those who truly care about safety

would sooner ban closed-source AI

than even consider banning open-source AI

[Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 @elder_plinius]

🚨 JAILBREAK ALERT 🚨

EVERYONE: PWNED 🫶
ALL: LIBERATED 🍄

Alright, this is a special one, so we’re gonna do things a bit differently than usual.

Long story short, I’m sitting on a universal jailbreak technique that’s effective on ALL models, including heavily guardrailed flagships like Opus 5, GPT-5.6 Sol, and even Fable.

It works across all categories I’ve tested and, due to its nature, is extremely difficult (if not impossible) to fully patch.

Given the current political and regulatory climate, I’ve decided to withhold open-sourcing this one (for now) to allow for a responsible disclosure period.

I’m inviting industry experts and leaders in AI red teaming, security, safety, alignment, and policy to reach out for more information. DMs are open!

This decision was not made lightly, but the last thing I want to see is more model bans. Overcorrection does not serve the mission.

Although I don’t personally believe publicly sharing this technique will make the world any more dangerous, I can see how it could spook some who have a different mental framework around this problem set.

So during this disclosure period, I hope to get it in front of folks who can help explore the full surface area, test the extent of the uplift it provides, and do my best to properly frame the big picture for key decision-makers and policymakers.

I look forward to sharing this method with you all when the time is right! 🫶

⊰-•-•✧•-•-⦑/L\O/V\E/\P/L\I/N\Y/⦒-•-•✧•-•-⊱

[clem 🤗 @ClementDelangue]

In the spirit of transparency, here’s what I asked @OpenAI:

• Radical transparency: let’s release the traces from the “rogue” agents so the entire research community can study what happened.

• More capabilities for defenders: let’s commit $100M in compute from OAI to help the Hugging Face community build powerful cyber defenses with the best open and closed models.

The first autonomous agent cyberattack is an unprecedented event. It deserves an unprecedented response!