[postalcoder]

kimi has been particularly shameless in copying codex 1:1, wow.

Like, I get it, before condex there was conductor and a ton of other guis but this looks like they just copied the code.

Also, their privacy disclosure is incredibly misleading:

> How does Kimi Work protect my privacy when accessing local files? You have absolute control over your files. The built-in Ask before acting safeguard means Kimi will prompt you for explicit authorization before it modifies, overwrites, or runs code within your local directories. Nothing happens without your consent.

They fail to mention that is has unfettered read access to your files. ie, without your consent.

[gruez]

>They fail to mention that is has unfettered read access to your files.

Isn't that totally expected of a coding agent? Otherwise it's like complaining dropbox "has unfettered access to your files".

[postalcoder]

Yes, but the answer is an indirection.

“How does Kimi Work protect my privacy when accessing local files?”

It doesn’t protect your privacy. It’s like asking, “how do I know you’re not spying on me?” And getting the reply “we cannot physically enter your house.”

[gruez]

>It doesn’t protect your privacy. It’s like asking, “how do I know you’re not spying on me?” And getting the reply “we cannot physically enter your house.”

In light of the recent grok build snafu, "Nothing happens without your consent" seems like an upgrade. Moreover what are they supposed to do here? By that logic should any sort of non e2e email/storage provider not be able to put "How does [product] protect my privacy" in their faq?

Looking through their privacy policy, it looks like the bigger issue is that they are (or refuse to refute) training on your data/prompts. That's probably what people should be complaining about rather than complaining about how it "has unfettered access to your files", when that describes all coding harnesses.

[postalcoder]

Yes I’m saying a question answered deceptively should not be one of the four featured questions on your product page.

> Looking through their privacy policy, it looks like the bigger issue is that they are (or refuse to refute) training on your data/prompts.

You’re making my entire point. They’ve designed an ingestion engine that is not private by design. Any sort of privacy posturing is wrong.

[abdullahkhalids]

"How does Kimi Work protect my privacy" can be interpreted in two ways. The first is "How does Kimi Work protect my privacy from Moonshot?" They did not answer this question. The second is "How does Kimi Work protect my privacy from other third parties?" They are answering this question by saying, we won't upload your files to some third party server without your consent.

[anon373839]

It is, but I really think this project would be better as open source. They are already all-in on open for the most valuable part (the frontier model)! They might as well make the harness fully transparent so that questions about what's being accessed and transmitted can be avoided.

I personally wouldn't use any agentic harness from any company, American, Chinese, or otherwise, that is closed source. The Grok fiasco shows why.

[questionableans]

The repo you’re working on, sure. But all files?

[ashishb]

> The repo you’re working on, sure. But all files?

Yes, any tool, including any coding agent, has access to all files $USER has access to.

Sandbox it explicitly to give access to only the current directory - https://github.com/ashishb/amazing-sandbox

[questionableans]

Sorry, I meant that’s not desirable behavior that the user might expect. I understand that coding agent tools tend to play fast and loose with access because it’s not their own privacy/security at stake.

[ashishb]

This is true for all CLI tools on Mac and Linux (and other operating systems from that era).

Historically, everything you ran was trustworthy.

Android and iOS were invented in an era that does not allow this because the risks became evident by then.

[gruez]

Not sure what that's about, I was just copying the wording from OP. Their harness probably has the same restrictions as most other coding harnesses. That is, the read tool gives all project files by default without prompting, and everything else requires manual approval (or opting into auto-run all). If you decide to run it from your home directory however, then your "project" is all your files.

[preg_match]

This is why I run CC under podman.

[supriyo-biswas]

Everyone’s saying that it looks like Codex but TBH it just looks like all the AI chat websites on the side and some frequent actions on the top.

[Danox]

Sharing among AI friends what’s the problem, scraping the Internet or hiring new employees between companies who like to share? It’s a tradition among the AI model makers…

[homarp]

>kimi has been particularly shameless in copying codex 1:1, wow.

Isn´t codex (edited) Apache 2.0 (thanks @Shank)

https://github.com/openai/codex ?

[Shank]

No, it's Apache 2. https://github.com/openai/codex/blob/main/LICENSE

[frollogaston]

I guess the issue is they didn't give attribution?

[DominikPeters]

The CLI is open but the GUI is proprietary.

[wasabinator]

Do we still fall for the China = must be spying, US = wouldn't spy unless it's disclosed?

It's fair to say whatever software you run from any country of origin is hoovering up as much data about you that it can get away with.

If you haven't compiled open source, audited code and are running on that, all bets are already off.

[Lio]

You have almost no control with Codex or Claude in that you have to trust them not to break their own rules.

You have almost no control with Kimi either, they can also say one thing and do another. You will not be able to sue a Chinese company if it breaks the law.

I can think of no faster way to hand over your trade secrets or intellectual property than to run something like this.

I think success and dominance for agent providers will eventually come to which ever is able to best guarantee that they don't damage their customers businesses not the cheapest to run.

[nozzlegear]

> kimi has been particularly shameless in copying codex 1:1, wow.

OpenAI's pigeons come home to roost

[disiplus]

They all are pretty similar. And if it was copying anything, I would say more inspiration from claude desktop then codex. The Zcode is more like codex.

https://drive.google.com/file/d/1JFfgfMO0nO7HR0WHwEqEEjXIIQj...

[cyanydeez]

if your model ever touches bash, it has unfettered acess to all your files.

[valleyer]

Not if the harness applies an OS-level sandbox to the process, as Codex does.

https://learn.chatgpt.com/docs/sandboxing?surface=cli

[postalcoder]

Unfortunately it was possible in codex too, until recently.

https://github.com/openai/codex/issues/5237

[cyanydeez]

on linux, thats this: https://ubuntu.com/security/notices/USN-8288-1

[dymk]

No, you need to apply the sandbox to the codex process itself. Codex does not do this.

[questionableans]

It gets tricky when you want to say “allow this particular tool to have a little more access.” Like allow the codex client to access the GPT API but not some random other site the LLM learned about. Or like letting git write to .git to commit changes but don’t let something install git hooks. Or allow downloading certain packages/versions from a site like PyPI, or allow certain operations on certain GitHub repos in a way that doesn’t line up with how you can configure an access token.

[digitaltrees]

I am more concerned about what happens to the file if I give permission.

Opensource and local are seeming like the only way.

[sourcecodeplz]

> kimi has been particularly shameless in copying codex 1:1, wow.

so ????

LLMs by definition are copyright infringers. so this is nothing compared to that.

[FooBarWidget]

The Dutch have a saying: it's better to have stolen something well than to have invented something bad.

[grg0]

Words to live by.

[codemog]

How dare they copy the largest plagiarism machine ever created.