this_user

Whenever people talk about why you should host in the EU, this is just the most recent example for why should shouldn't.

Nuclear is to solar in terms of energy generation what hydrogen is to electric for cars: it's proponents keep arguing that the other thing doesn't work and will never work, and that you absolutely, positively need their approach. And then it just works and effectively kills off the other technology within a couple of years. Nuclear is basically already dead, and anyone starting new nuclear projects now will look extremely silly in 20-30 years when they finally come online.

That is how you end up with a test suite that now takes 10x as long, unless you occasionally prompt the agent to consolidate and group tests.

There are still a lot of issues with coding that require intervention. Just recently, I had Astra detect a minor bug in an HTTP client that resulted in it deciding that the best approach would be to spend 1800 LOC on implementing its own from scratch. Some other time, Opus 5.5 was trying to tell me that test runs on a small to mid-sized application would take another three weeks until they would be fully finished.

They are probably like 3-6 months away from just letting their agents work directly on prod after realising that repos don't scale enough.

If you need a reason, the modem and the baseband firmware have several unpatched vulnerabilities on that model, which is not something that an Android update can resolve.

OpenSSH don't have the same luxury of being able to ignore potential vulnerabilities.

Well, the point is being able to charge enterprise customers more.

And how many CDs can you buy per month for the price of a Spotify subscription? I am not going to be spending hundreds of Euros per month on physical media, and I also value my time too much to bother with ripping and converting these, just so I can end up with a worse experience for more money, while also producing a bunch of unnecessary plastic rubbish.

Astra doesn't just burn token at an insane rate, it is also strangely high maintenance when using it. Occasionally, you have to keep prodding it to keep working. Then at other times, it will disappear down some rabbit hole, trying to resolve increasingly hypothetical issues. It feels like you constantly have to keep it on track, while Opus is just churning through tasks.

They probably forgot to tell Claude to make no mistakes.

But seriously, the fact that this started as Crypto-adjacent should have immediately disqualified them for serious use.

Astra is barely usable even on the $100 plan. And that is if it doesn't just burn through 80% of your weekly quota in a couple of hours by continually expanding the scope of the task you gave it - while not noticing the failing tests that are right in front of it.

Opus is at least actually usable even on the small plan. The main downside is its insane writing style, but 5.5 seems to address that somewhat. Otherwise, you can just use your $20 OpenAI plan to have Luna de-slop Opus' prose, which seems to work fine.

The issue is that in order to have the agent write good code, you need to implement standard SWE best practices. But that also means a lot of manual intervention in terms of writing specs, checking acceptance criteria, and reviewing code. So you end up spending a lot of time on managing your agent, which means you won't get a 1000% productivity gain, you get maybe 50 or 100, possible less in some areas and with some issues.

I'm not sure that Claude's "Realigned the shape of the load-bearing ownership gate to reduce the blast radius of the design contract; confirmed, not assumed" is more meaningful than "fix".

Welcome to agentic coding in 2026.

They are still making billions from highly sticky enterprise customers, which is more than enough to survive. Otherwise, Microsoft would have gone out of business years ago.

No need for QA if you just tell Claude to "make no mistakes".

Now their broken slop code is at least memory safe.

Nothing about this is surprising, something like this was always going to happen, because you - or an LLM for that matter - can always find a line of motivated reasoning that justifies any course of action. One would have to be extremely naive to believe that "alignment" provides any kind of actually robust guardrails. Simultaneously, we have seen decades of security vulnerabilities. Unless your testbed is truly and fully physically airgapped, any current SOTA model will find a way to break out.

The reality is that the current approach to AI safety is little more than a fig leaf, but you also won't be able to put the genie back in the bottle, because the technology is simply too powerful to abandon. There is always going to be someone developing it further from now on.

So, the real question is what a novel and actually effective approach to AI safety looks like and how to get there.

It's honestly a smart decision. They already have access to the common market just like Norway. Becoming a full member just means being subject to a EU governing body that has proven itself to be largely ineffective and incompetent, sometimes reaching the realm of "outright harmful" with things like Chat Control. Why would you want to be a part of that, especially if you would also be paying for that privilege?

You should probably realise that the world has radically changed since then. This kind of thing works when you have a significant lead in the field that makes keeping vulnerabilities open sufficiently low risk for your own side. But if your adversaries have similar capabilities, then the calculation changes.

Would be interesting to compare the Chinese version. Because, obviously, their English version is for users, not for investors or government officials, while the US labs are always addressing those too.

That always seems to happen upon release of a new model. People look at the benchmarks, which look good, because the model was almost certainly optimised for that. Then they start actually using it, and after a week or two we get the real assessment that is almost always less impressive than the initial reactions.

This is not even news. This piece of information was already included in the original release, and the issue was apparently a miscommunication between the two companies that left a gap open in the test environment.

Instead of rewriting Bun, maybe Anthropic should have thrown some tokens at finishing Hurd.

I am not sure what point the author is even trying to make here. On the one hand, he seems to complain about how AI has virtually made the traditionally PhD thesis obsolete, but on the other hand, he also states:

> Consider that most PhD theses were never good. How often do you rush to read a PhD thesis? The vast majority of them are painful to read. You learn little if anything.

So, it sounds like nothing of much value has been lost.

I think what he really complains about is that AI is starting to show that the emperor called academia has no clothes. So much of working your way through that system has always been about being able to master largely pointless rituals.

Yet, the people on the inside have no interest in making any improvements, because academia has always been institutionally conservative. But now AI is starting to put pressure on them to rethink their way of doing things, and they really don't like it.

Not really though, since the result of the prompt is not deterministic. It greatly depends on the model, the version, the harness, even time of day if the provider's infrastructure is currently overloaded and is silently degrading performance. Some things also require multi-turn interactions.

Almost Microsoft material.

This had nothing much to do with AI in particular though. The issue was that there had previously been a military installation at the same location, but the US military did not update its database. Apparently, one analyst had flagged the issue for review in 2019, but that was never done before the strike.

The point is that the same thing could have happened just as easily if you had selected targets based on the faulty data by any other means than an AI model.

They are, which is why this pearl clutching over particles from cookware is a bit silly.

> China wants the US economy to flounder.

Why would they possibly want their largest customer to flounder?