[jryio]

It's so important to remember that unlike code which can be reverted - most file system and application operations cannot.

There's no sandboxing snapshot in revision history, rollbacks, or anything.

I expect to see many stories from parents, non-technical colleagues, and students who irreparably ruined their computer.

Edit: most comments are focused on pointing out that version control & file system snapshot exists: that's wonderful, but Claude Cowork does not use it.

For those of us who have built real systems at low levels I think the alarm bells go off seeing a tool like this - particularly one targeted at non-technical users

[Workaccount2]

Frequency vs. convenience will determine how big of a deal this is in practice.

Cars have plenty of horror stories associated with them, but convenience keeps most people happily driving everyday without a second thought.

Google can quarantine your life with an account ban, but plenty of people still use gmail for everything despite the stories.

So even if Claude cowork can go off the rails and turn your digital life upside down, as long as the stories are just online or "friend of a friend of a friend", people won't care much.

[soiltype]

Considering the ubiquity and necessity of driving cars is overwhelmingly a result of intentional policy choices irrespective of what people wanted or was good for the public interest... actually that's quite a decent analogy for integrated LLM assistants.

People will use AI because other options keep getting worse and because it keeps getting harder to avoid using it. I don't think it's fair to characterize that as convenience though, personally. Like with cars, many people will be well aware of the negative externalities, the risk of harm to themselves, and the lack of personal agency caused by this tool and still use it because avoiding it will become costly to their everyday life.

I think of convenience as something that is a "bonus" on top of normal life typically. Something that becomes mandatory to avoid being left out of society no longer counts.

[yencabulator]

I mean, we were there before this Cowork feature started exposing more users to the slot machine:

"Claude CLI deleted my home directory and wiped my Mac" https://news.ycombinator.com/item?id=46268222

"Vibe coding service Replit deleted production database, faked data, told fibs" https://news.ycombinator.com/item?id=44632575

"Google Antigravity just deleted the contents of whole drive" https://news.ycombinator.com/item?id=46103532

[Quothling]

> So even if Claude cowork can go off the rails and turn your digital life upside down, as long as the stories are just online or "friend of a friend of a friend", people won't care much.

This is anecdotal but "people" care quite a lot in the energy sector. I've helped build our own AI Agent pool and roll it out to our employees. It's basically a librechat with our in-house models, where people can easily setup base instruction sets and name their AI's funny things, but are otherwise similar to using claude or chatgpt in a browser.

I'm not sure we're ever going to allow AI's access to filesystems, we barely allow people access to their own files as it is. Nothing that has happened in the past year has altered the way our C level view the security issues with AI in any other direction than being more restrictive. I imagine any business that cares about security (or is forced to care by leglislation) isn't looking at this as a they do cars. You'd have to be very unlucky (or lucky?) to shut down the entire power grid of Europe with a car. You could basically do it with a well placed AI attack.

Ironically, you could just hack the physical components which probably haven't had their firmware updated for 20 years. If you even need to hack it, because a lot of it frankly has build in backdoors. That's a different story that nobody on the C levels care about though.

[alwillis]

The first version is for macOS, which has snapshots [1] and file versioning [2] built-in.

[1]: https://eclecticlight.co/2024/04/08/apfs-snapshots/

[2]: https://eclecticlight.co/2021/09/04/explainer-the-macos-vers...

[shepherdjerred]

Are average users likely to be using these features? Most devs at my company don’t even have Time Machine backups

[cbm-vic-20]

RSX-11M for the PDP-11 had filesystem versioning back in the early 1980s, if not earlier.

[falcor84]

Once upon a time, in the magical days of Windows 7, we had the Volume Shadow Copy Service (aka "Previous Versions") available by default, and it was so nice. I'm not using Windows anymore, and at least part of the reason is that it's just objectively less feature complete than it used to be 15 years ago.

[superjose]

Yeah. I also like Windows, but MS has done a wonderful job to destroy the OS with newer releases.

I haven't had to tweak an OS like Win 11 ever.

[hopelite]

Somewhat related is a concern I have in general as things get more "agentic" and related to the prompt injection concerns; without something like legally bullet-proof contracts, aren't we moving into territory of basically "employing" what could basically be "spies" at all levels from personal (i.e., AI company staff having access to your personal data/prompts/chats) to business/corporate espionage, to domestic and international state level actors who would also love to know what you are working on and what you are thinking/chatting about and maybe what your mental health challenges are that you are working through with an AI chat therapist.

I am not even certain if this issue can be solved since you are sending your prompts and activities to "someone else's computer", but I suspect if it is overlooked or hand-waved as insignificant, there will be a time when open, local models will become useful enough to allow most to jettison cloud AI providers.

I don't know about everyone else, but I am not at all confident in allowing access and sending my data to some AI company that may just do a rug pull once they have an actual virtual version of your mind in a kind of AI replication.

I'll just leave it at that point and not even go into the ramifications of that, e.g., "cybercrimes" being committed by "you", which is really the AI impersonator built based on everything you have told it and provide access to.

[toddmorey]

Q: What would prevent them from using git style version control under the hood? User doesn’t have to understand git, Claude can use it for its own purposes.

[twosdai]

Didn't actually check out the app, but some aspects of application state are hard to serialize, some operations are not reversible by the application. EG: sending an email. It doesn't seem naively trivial to accomplish this, for all apps.

So maybe on some apps, but "all" is a difficult thing.

[nikkwong]

You can’t easily snapshot the current state of an OS and restore to that state like with git.

[samuelstros]

Git only works for text files. Everything else is a binary blob which, among other things, leads to merge conflicts, storage explosion, and slow git operations

[y42]

Indeed there are and this is no rocket science. Like Word Documents offer a change history, deleted files go to the trash first, there are undo functions, TimeMachine on MacOs, similar features on Windows, even sandbox features.

[fuzzy2]

Trash is a shell feature. Unless a program explicitly "moves to trash", deleting is final. Same for Word documents.

So, no, there is no undo in general. There could be under certain circumstances for certain things.

[cush]

State isn't always local too

[kamaal]

>>I expect to see many stories from parents, non-technical colleagues, and students who irreparably ruined their computer.

I do believe the approach Apple is taking is the right way when it comes to user facing AI.

You need to reduce AI to being an appliance that does one or at most a few things perfectly right without many controls with unexpected consequences.

Real fun is robots. Not sure no one is hurrying up on that end.

>>Edit: most comments are focused on pointing out that version control & file system snapshot exists: that's wonderful, but Claude Cowork does not use it.

Also in my experience this creates all kinds of other issues. Like going back up a tree creates all kinds of confusions and keeps the system inconsistent with regards to whatever else it is you are doing.

You are right in your analysis that many people are going to end up with totally broken systems

[bob1029]

In theory the risk is immense and incalculable, but in practice I've never found any real danger. I've run wide open powershell with an OAI agent and just walked away for a few hours. It's a bit of a rush at first but then you realize it's never going to do anything crazy.

The base model itself is biased away from actions that would lead to large scale destruction. Compound over time and you probably never get anywhere too scary.

[seunosewa]

There's no reason why Claude can't use git to manage the folders that it controls.

[binarymax]

Most of these files are binary and are not a good fit for git’s graph based diff tracker…you’re basically ending up with a new full sized binary for every file version. It works from a version perspective, but is very inefficient and not what git was built for.

[oblio]

Git isn't good with big files.

I wanted to comment more, but this new tool is Mac only for now, so there isn't much of a point.

[Weryj]

TimeMachine has never been so important.

[fragmede]

Arq does it better.

[greenavocado]

TimeMachine is worthless trash compared to restic

[hans0l074]

IIUC, this is a preview for Claude Max subscribers - I'm not sure we'll find many teachers or students there (unless institutions are offering Max-level enterprise/team subscriptions to such groups). I speculate that most of those who will bother to try this out will be software engineering people. And perhaps they will strengthen this after enough feedback and use cases?

[Aeolun]

If this is like Claude Code for everyone else, shouldn’t it be snapshotting anything it changes so that you can go back to the previous state?

[machiaweliczny]

Yeah, seems like this could be achieved by using https://github.com/streamich/memfs/blob/master/docs/snapshot...

Weird they don't use it - might backfire hard

[matt3D]

Pretty much every company I work with uses the desktop sync tools for OneDrive/GoogleDrive/Dropbox etc.

It would be madness to work completely offline these days, and all of these systems have version history and document recovery built in.

[__MatrixMan__]

I hope we see further exploration into immutable/versioned filesystems and databases where we can really let these things go nuts, commit the parts we want to keep, and revert the rest for the next iteration.

[Helmut10001]

I would never use what is proposed by OP. But, in any case, Linux on ZFS that is automatically snapshotted every minute might be (part of) a solution to this dilemma.

[akurilin]

You make a good point. I imagine that they will eventually add Perforce-style versioning to the product and this issue will be solved.

[o_m]

So the future is NixOS for non-technical people?

[porkloin]

Yes, and I think we're already seeing that in the general trend of recent linux work toward atomic updates. [bootc](https://developers.redhat.com/articles/2024/09/24/bootc-gett...) based images are getting a ton of traction. [universal blue](https://universal-blue.org/) is probably a better brochure example of how bootc can make systems more resilient without needing to move to declarative nix for the entire system like you do in NixOS. Every "upgrade" is a container deployment, and you can roll back or forward to new images at any time. Parts of the filesystem aren't writeable (which pisses people off who don't understand the benefit) but the advantages for security (isolating more stuff to user space by necessity) and stability (wedged upgrades are almost always recoverable) are totally worth it.

On the user side, I could easily see [systemd-homed](https://fedoramagazine.org/unlocking-the-future-of-user-mana...) evolving into a system that allows snapshotting/roll forward/roll back on encrypted backups of your home dir that can be mounted using systemd-homed to interface with the system for UID/GID etc.

These are just two projects that I happen to be interested in at the moment - there's a pretty big groundswell in Linux atm toward a model that resembles (and honestly even exceeds) what NixOS does in terms of recoverability on upgrade.

[teekert]

Or rather ZFS/BTRFS/BchachFS. Before doing anything big I make snapshot, saved me recently when a huge Immich import created a mess, `zfs rollback /home/me@2026-01-12`... And it's like nothing ever happened.

[big-chungus4]

A human can also accidentally delete or mess up some files. The question is whether Claude Cowork is more prone to it.

[heliumtera]

There was a couple of posts here on hacker news praising agents because, it seems, they are really good at being a sysadmin. You don't need to be a non-technical user to be utterly fucked by AI.

[TeMPOraL]

Theoretically, the power drill you're using can spontaneously explode, too. It's very unlikely, but possible - and then it's much more likely you'll hurt yourself or destroy your work if you aren't being careful and didn't set your work environment right.

The key for using AI for sysadmin is the same as with operating a power drill: pay at least minimum attention, and arrange things so in the event of a problem, you can easily recover from the damage.

[neocron]

Not a big problem to make snapshots with lvm or zfs and others. I use it automatically on every update

[lp0_on_fire]

What percentage of non-IT professionals know what zfs/lvm are let alone how to use them to make snapshots?

[fouronnes3]

I'm not even sure if this is a sarcastic dropbox-style comment at this point.