Komlan Egoh

[Poonam Soni @CodeByPoonam]

Do you understand what the US government just did to Anthropic?

They forced them to pull Claude's most powerful AI model today.

Claude Fable 5 and Mythos 5 are gone. Both shut down by export control directive.

Anthropic received the letter at 5:21pm ET. By the end of the day, the most capable AI model on earth was no longer accessible to anyone. Not to enterprise customers paying millions. Not to developers building on the API. Not to the hundreds of millions of people using Claude.

The reason: the government claims someone found a way to jailbreak Fable 5. The jailbreak in question? Asking the model to read a codebase and find software flaws. Anthropic's response: that's something every defender does every day. GPT-5.5 does it too. They published the receipts.

Anthropic spent thousands of hours red-teaming Fable 5 alongside the US government, the UK AISI, and outside organizations before launch. The same government that helped test it ordered them to pull it.

The most telling line came from Anthropic's own statement: "We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers."

They're complying anyway.
AI policy stopped being theoretical at 5:21pm ET today.

This is the first time a frontier model has been pulled by government order. It will not be the last.

[Komlan Egoh @KomlanEgoh]

@AnthropicAI When will access be restored to us nationals? How will you ensure any individual customer is a us national?

[Andrej Karpathy @karpathy]

This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc.

More generally, imo audio is the human-preferred input to AIs but vision (images/animations/video) is the preferred output from them. Around a ~third of our brains are a massively parallel processor dedicated to vision, it is the 10-lane superhighway of information into brain. As AI improves, I think we'll see a progression that takes advantage:

1) raw text (hard/effortful to read)
2) markdown (bold, italic, headings, tables, a bit easier on the eyes) <-- current default
3) HTML (still procedural with underlying code, but a lot more flexibility on the graphics, layout, even interactivity) <-- early but forming new good default
...4,5,6,...
n) interactive neural videos/simulations

Imo the extrapolation (though the technology doesn't exist just yet) ends in some kind of interactive videos generated directly by a diffusion neural net. Many open questions as to how exact/procedural "Software 1.0" artifacts (e.g. interactive simulations) may be woven together with neural artifacts (diffusion grids), but generally something in the direction of the recently viral https://t.co/z21CP5iQfu

There are also improvements necessary and pending at the input. Audio nor text nor video alone are not enough, e.g. I feel a need to point/gesture to things on the screen, similar to all the things you would do with a person physically next to you and your computer screen.

TLDR The input/output mind meld between humans and AIs is ongoing and there is a lot of work to do and significant progress to be made, way before jumping all the way into neuralink-esque BCIs and all that. For what's worth exploring at the current stage, hot tip try ask for HTML.

[Komlan Egoh @KomlanEgoh]

This is absolutely true at the moment, but we can expect the models to get better and make les mistakes.

[Louie Bacaj @LBacaj]

Look, I used to manage software engineers. I was always very conscious about over hiring because engineers cost a lot.

And now I am doing a lot of AI aided coding.

I can promise you I haven’t seen a single thing that tells me we won’t need software engineers in the future.

It’s the exact opposite.

These models are great at speeding up software engineering work, but they still
make a lot of mistakes. The moment the complexity of the code base reaches above a certain level, they fall apart without targeted guidance by someone who understands code.

You might say, “ok but if engineers can move faster, won’t we need less software engineers?” Maybe. But I really don’t think so because the amount of code out there is also about to explode.

[Min Choi @minchoi]

MCP is going crazy viral right now🤯

AI apps can now instantly connect to any tool or live data.

USB-C moment for AI.

10 wild examples:

[Komlan Egoh @KomlanEgoh]

@AlexHormozi So true!

A focused fool can accomplish more than a distracted genius.

[Martin Fowler @martinfowler]

NEW POST

DeepSeek's LLMs made a big splash, but more interesting is their recent research papers. Shayan Mohanty writes an overview of them, outlining their three main arcs: efficiency, HPC Co-Design, and RL for emergent reasoning.

https://t.co/DdBxRETrYD

[Martin Fowler @martinfowler]

Much of what I'm reading about the KOSA bill in the U.S. sounds worrying. Here is a good example of the concerns

https://t.co/pM8ebJpM1L

[Komlan Egoh @KomlanEgoh]

@ElonMuskAOC @POTUS What's your point?

[Nikki Siapno @NikkiSiapno]

These 7 GitHub repositories will take your Python skills to the next level:

[Chris Hayes @chrislhayes]

As parents, we're at the age where we are truly entering The Organized Sports era and I guess we just...don't get to do anything other than take kids to games and practices for the next 6-8 years. Is that the deal?

[Greg Young @gregyoung]

This made me LOL way too much.

[Komlan Egoh @KomlanEgoh]

Wonder!

[NASA JPL @NASAJPL]

A red-letter day on the Red Planet! #MarsHelicopter

[Joe Scarborough @ScarboroughNow]

Amy, Mika and I are so sorry to hear about your terrible loss. We are also grateful you reached out to tell us about Jane. We are blessed to have had such a wonderful member of our TV family.

[Programming Wisdom @CodeWisdom]

"The first step of any project is to grossly underestimate its complexity and difficulty." – Nicoll Hunt

[Steve Stewart-Williams @SteveStuWill]

This photo of the sun might not look too impressive... until you realize it was taken at night – not looking up but looking down, through the entire Earth, using neutrinos rather than light. Amazing! https://t.co/GiMKv9NQ7E

[Komlan Egoh @KomlanEgoh]

Very good work @rockthejvm in 2020. Keep up the good work. https://t.co/3QSlEKZyb5

[Komlan Egoh @KomlanEgoh]

‘It will change everything’: DeepMind’s AI makes gigantic leap in solving protein structures https://t.co/6J5ssUodmn

[Komlan Egoh @KomlanEgoh]

@visseho09 @UCT_news @UQAM @MPIDRnews Congrats @visseho09. I always knew you were going places.

[Komlan Egoh @KomlanEgoh]

@BenSasse Since when has kneeling become a sign of disrespect?