I would have agreed with this like 15 years ago, but the very existence of Twitter (and the acquisition saga) proves this to not be true.

But in the real world, the systems that deal with processing people's entries already cross-reference multiple other existing databases, require internet connectivity to do so, and I think you'll have hard time convincing anyone to stop doing that.

The internet requirement is not there for the person presenting the document, it's for the person/system checking it.

If CBP's systems go down, they will not process (foreign, they'll process US citizens still) arrivals [1], even with physical passports in front of them. I assume the EU ESS works the same.

"If the internet goes down, your border checkpoint is down" is not some terrifying future we need to protect against, it's the reality of the world as you live in right now.

[1]: I've had to wait for an hour, at SFO of all places, because of exactly that happening.

You can use Google to find you results reinforcing your belief that the earth is flat too; but we don't condemn Google as a helpful tool during research.

If you trust whatever the LLM spits out unconditionally, that's sorta on you. But they _can_ be helpful when treated as research assistants, not as oracles.

You still need to verify it, but "find the right things to read in the first place" is often a time intensive process in itself.

(You might, at that point, argue that "what if LLM fails to find a key article/paper/whatever", which I think is both a reasonable worry, and an unreasonable standard to apply. "What if your google search doesn't return it" is an obvious counterpoint, and I don't think you can make a reasonable argument that you journalists should be forced to cross-compare SERPs from Google/Bing/DuckDuckGo/AltaVista or whatever.)

Wow, such a risky bet, I'm not sure it'll pay off.

I Am Not A Doctor And This Is Not Medical Advice.

(I think?).

This wasn't big enough in Tokyo to send out one.

I think you're right to be skeptical, but they _have_ talked about the process publicly.

And I don't think there's anything there that is not reproducible by outsiders? They have access to the same Opus 4.6 that you and I do; though not having to pay for the tokens certainly helps.

I'm pretty sure if you wanted to burn a couple thousand bucks, you'd reproduce at least some of these findings.

They also say publicly in their Opus 4.6 post (https://red.anthropic.com/2026/zero-days/):

>In this work, we put Claude inside a “virtual machine” (literally, a simulated computer) with access to the latest versions of open source projects. We gave it standard utilities (e.g., the standard coreutils or Python) and vulnerability analysis tools (e.g., debuggers or fuzzers), but we didn’t provide any special instructions on how to use these tools, nor did we provide a custom harness that would have given it specialized knowledge about how to better find vulnerabilities. This means we were directly testing Claude’s “out-of-the-box” capabilities, relying solely on the fact that modern large language models are generally-capable agents that can already reason about how to best make use of the tools available.

Maybe bundling these under the same system is a mistake and they should be separate systems with different considerations; it would certainly help with arguments about it online ;P

- - -

Hey, I tried restoring a file from my backup — downloading it directly didn't work, and creating a restore with it also failed – I got an email telling me contract y'all about it.

Can you explain to me what happened here, and what can I do to get my file(s?) back?

- - -

Hi Jan,

Thanks for writing in!

I've reached out to our engineers regarding your restore, and I will get back to you as soon as I have an update. For now, I will keep the ticket open.

- - -

Hi Jan,

Regarding the file itself - it was deleted back in 2022, but unfortunately, the deletion never got recorded properly, which made it seem like the file still existed.

Thus, when you tried to restore it, the restoration failed, as the file doesn't actually exist anymore. In this case, it shouldn't have been shown in the first place.

For that, I do apologize. As compensation, we've granted you 3 monthly backup credits which will apply on your next renewal. Please let me know if you have any further questions.

- - -

That makes me even more confused to be honest - I’ve been paying for forever history since January 2022 according to my invoices?

Do you know how/when exactly it got deleted?

- - -

Hi Jan,

Unfortunately, we don't have that information available to us. Again, I do apologize.

- - -

I really don’t want to be rude, but that seems like a very serious issue to me and I’m not satisfied with that response.

If I’m paying for a forever backup, I expect it to be forever - and if some file got deleted even despite me paying for the “keep my file history forever” option, “oh whoops sorry our bad but we don’t have any more info” is really not a satisfactory answer.

I don’t hold it against _you_ personally, but I really need to know more about what happened here - if this file got randomly disappeared, how am I supposed to trust the reliability of anything else that’s supposed to be safely backed up?

- - -

Hi Jan,

I'll inquire with our engineers tomorrow when they're back in, and I'll update you as soon as I can. For now, I will keep the ticket open.

- - -

Appreciate that, thank you! It’s fine if the investigation takes longer, but I just want to get to the bottom of what happened here :)

- - -

Hi Jan,

Thanks for your patience.

According to our engineers and my management team:

With the way our program logs information, we don't have the specific information that explains exactly why the file was removed from the backup. Our more recent versions of the client, however, have vastly improved our consistency checks and introduced additional protections and audits to ensure complete reliability from an active backup.

Looking at your account, I do see that your backup is currently not active, so I recommend running the Backblaze installer over your current installation to repair it, and inherit your original backup state so that our updates can check your backup.

I do apologize, and I know it's not an ideal answer, but unfortunately, that is the extent of what we can tell you about what has happened.

- - -

I gave up escalating at this point and just decided these aren’t trusted anymore.

The files in question are four year old at this point so it’s hard for me conclusively state, so I guess there might be a perfect storm of that specific file being deleted because it was due to expire before upgraded to “keep history forever”, but I don’t think it’s super likely, and I absolutely would expect them to have telemetry about that in any case.

If anyone from Backblaze stumbles upon it and wants to escalate/reinvestigate, the support ID is #1181161.

I contacted the support asking WTF, "oh the file got deleted at some point, sorry for that", and they offered me 3 months of credits.

I do not trust my Backblaze backups anymore.

To _their phone_ specifically? Probably almost nobody. But to their Google/Apple Photos library?

A lot, if not most of people who use DSLRs and other point-and-shoot cameras. Most people want a single library of photos, not segregated based on which device they shot it on.