Altman was fired by his own board for lying to them. Just because Microsoft blackmailed them into reversing this decision by threatening financial ruin does not change that.

You don't give habitual liars the benefit of doubt.

If you look at his comments about Palantir and their proposed safeguards, it's clear it's a case of "if you are dining with the Devil, you'd better bring a very long spoon"

These comments were after the deal had soured. Not before. If it was a case of such morality, the partnership with Palantir would have never happened in the first place.

The contract was explicit - it was for defence purposes with a company known for spying activities. So, obviously spying is involved and they weren't just going to generate cat videos with it.

Again, nobody is innocent here.

Or, as is likely, OpenAI models have no guardrails, Anthropic's did and the DoD was bumping into them.

Does anyone else notice claude is just plain better at reasoning? It may not just be post training guardrails. It would not surprise me of it was something anthropic couldn't simply disable. Either from reinforcement or even training corpus curation. Of all the models, claude is the only one that makes me wonder if they have figured out something beyond stochastic language generation and aren't telling anyone

I have noticed this too, despite the close benchmark results Claude just works better. It knows when to push back, it has an "agency"... there is something there that I don't see with Gemini or OpenAI's best paid models.

Thanks for the writeup, Ivan, I am a great fan of your work!

Now we need to get Qualys to cap SSL Labs ratings at B for servers that don't support ECH. Also those that don't have HSTS and HSTS Preload while we're at it.

Thanks! Sadly, SSL Labs doesn't appear to be actively maintained. I've noticed increasing gaps in its coverage and inspection quality. I left quite a while ago (2016) and can't influence its grading any more, sadly.

Is there a well-maintained alternative to SSL Labs you can recommend?

Yes, there is! After I left SSL Labs, I built Hardenize, which was an attempt to go wider and handle more of network configuration, not just TLS and PKI. It covers a range of standards, from DNS, over email, TLS and PKI, and application security.

Although Hardenize was a commercial product (it was acquired in 2022 by another company, Red Sift), it has a public report that's always been free. For example:

https://www.hardenize.com/report/feistyduck.com

The CSP inspection in Hardenize could use a refresh, but the TLS and PKI aspects are well maintained [at the time of writing].

I use testssl.sh [1] mostly because I can test things not publicly accessible.

[1] - https://github.com/testssl/testssl.sh

I am coming to GrapheneOS from iOS, as recounted in https://blog.majid.info/quit-apple/

My experience with seedvault is not as positive as the OP, it fails for about half my apps. I guess I have another year to fix this when the GrapheneOS enabled Motorola flagship comes out.

I carry both my iPhone 16 Plus and my Pixel 8 Pro, but limit my iPhone use to mobile banking, Apple Pay and Find My.

Seedvault is godawful and I'm saying it as a someone using GOS as a daily driver. Utterly awful.

Two backup runs can give different results (failing in different places or failing to restore everything), and I haven't heard of the one person it works for every time.

I wish I could just run Titanium Backup :(

Log into webapps,

sync files,

new phone set up.

Did you read the article? Apple's servers are M2 Ultra class and not able to run modern models.

What does that matter? They cant be reused for other things? Nonsense.

What other services does Apple have that people would be paying for? The ones they have today are either iCloud storage, which does not need much compute, or merely an alibi so they can claim with an almost straight face that Apple's "Services" revenue isn't basically just the App Store 30% tax. That also explains why they are constantly shoving ads for News or Fitness our throats in the Settings app.

And no other services will exist or grow?

Using something like Taalas' hardcoded model as opposed to running one on general purpose GPUs, flexible but power-hungry.

https://www.cnx-software.com/2026/02/22/taalas-hc1-hardwired...

And Weasyprint does not have browser dependencies either, which is great.

Fun fact: I had to write a routine administrative letter for my parents in another country, I asked Claude to do so in PDF form so I could email it to them they would print it and mail it. The way it did so was to write a Python program using Weasyprint to generate the PDF...

Overly enthusiastic Rust evangelists can be annoying, but nowhere as much as C++ or C advocates defensively claiming memory safety isn't a big deal, and they are going to have it in the next version of the language anyway.

I find my experience with Erlang has helped with the (considerable) learning curve for Rust, but I still prefer Go for most use-cases.

> claiming memory safety isn't a big deal

There are contexts where it is, there are contexts where it is not.

But suddenly everyone out there is dealing only with those context where it is.