Kimi Work attaches raw agent sessions to feedback reports | Hacker News Reader

Kimi Work attaches raw agent sessions to feedback reports

20 pointsryanmerket8 comments

I reverse engineered the latest Kimi Work desktop app just for fun, I know I'm a nerd.

Anyway, I found something a bit concerning: when you send a feedback report, it attaches the 5 latest sessions to the report with no notice or heads up or anything.

These sessions could have ANYTHING in them, and you have no idea you're sending it all to Kimi.

I emailed them to give them a heads up, but just wanted to let yall know incase you're using Kimi.

URL: https://runtimewire.com/article/scoop-kimi-work-secretly-att...

View on HN

> Why It Matters

Thanks Claude, the analysis looks to be load-bearing, and dare I say the gun may be smoking. This sharpens the framing considerably.

Honest caveat: most AI providers' privacy policies have an explicit carve-out for feedback. (Uploading the last 5 sessions wholesale is pretty wild though.)

FWIW, this particular header in news articles is a deliberate choice originated by Axios. It’s notable enough and effective enough they wrote a book about it.

https://www.axios.com/smart-brevity

Did Claude write this article? Probably. But this style of article is exactly what you’d expect from a pre-AI version of this website as well.

It goes back way earlier than Axios.

You’ve submitted nearly 100 AI written articles from this source in the past 40 days…

Yes and I'm having a blast, being semi-retired and reporting the news before anyone esle. Heck, even Wired covered my insane levels of reporting.

AI has completely killed blogs for me, I don't bother anymore unless I'm familiar with the blog from years before.

I feel the same way but don’t let some bad blogs dissuade you from others IMO. You can tell pretty quickly if the author cares enough to write it themselves or have an agent do it.

> when you send a feedback report, it attaches the 5 latest sessions to the report with no notice or heads up or anything.

Not informing the user of that feels wrong, and even then you'd probably want like a dropdown to choose between just the current session vs the last 5 if that much context is really needed, with either Off or 1 being the default.

Claude Code has a clear disclaimer, probably exactly how it should be done:

> Submitting this report will send the entire current conversation to Anthropic for future improvements to our models. Learn More

Reply on news.ycombinator.com