Many people don't wanna depend on US products. If US switches them off, you are gone. Not even talking about privacy of US companies. Nor talking about guardrails that let you not do things you need to do.
1) What kind of guardrails are you using? Do you do models like prompt injection checkers? Or what guardrails are useful that do not know the environment the agent lives in?
2) I heard Bifrost also works for this. Did you try that as well, pros/cons?
I'm not the person you asked but yes pretty much. I've found the prompt injection protection models pretty limited, but get good results from gpt-oss-safeguard with a policy describing various prompt injection techniques. More expensive but more flexible and reliable and I only run it on web fetch and web search results. Costs me like 5-20 cents a day depending on what I'm working on.
Still working on deploying it and setting it up with cloudflare sandboxes so I don't have deep expertise and need to experiment first before I can give advice
Labs typically pay $2k for each [prompt+scorer] docker image. So this is why frontier labs need so much cash and manpower and they see it as their moat.
It also means security is not held as high and vulnerabilities not as much found. A simple 0-day may survive for years. Not much effort needed to have permanent access.
Its easier for an LLM to find vulnerabilities in a project with less usage, and those vulnerabilities will stay open longer, making them much cheaper to attack to keep the door open.
What's the point of attacking projects that almost nobody uses?
Do you think Netanyahu, Trump or Xi-Jinping are somehow secretly using Cosmic DE at home, to be worthy targets?
Bad actors have limited time, lives of their own and mouths to feed as well, so they concentrate their efforts where "the fish are" if they want to PWN someone for profit.
That's why Windows was the biggest target in the past for so long and why MacOS and Linux were ignored. Because most of the fish were on Windows.
Offensive security employees, tokens, and peoples' time are still a finite resource that get allocated based on target priorities and operational end-goals, even by state actors.
If you assume Mossad and NSA are Token-maxxing every single niche FOSS project out there to cast as large as possible fishnet on hacking all Average Joes on the planet just in case, then maybe using Mozilla and MacOS gets you hacked too, maybe even visiting HN and commenting here gets you hacked by some zero days you don't yet know.
No. Prefix caching is just an optimisation on the server side. What Kimi CLI does is insert <system> tags that include a checkpoint marker with an id.
The model is then given a tool that allows the model to decide to roll back to a checkpoint + a message containing any additional useful information.
It's specifically instructed to use that tool[1] in cases like when it has inadvertendly read a large file where most of the content is not relevant to the task, or after a web search where it's found what it's looking for but most of the content isn't needed, or when it's written code that didn't work as expected, or similar.
It basically lets the model backtrack and "forget" irrelevant details at the end of the context but give itself hints on how it should continue from the checkpoint.
Though, interestingly they seem to be abandoning it in their new CLI (kimi-code), unless it's been folded into other functionality. Not sure if they just feel it's not needed any more with their newer models or if it just didn't work as well as they expected.
Prefix caching automatically detects priorly read tokens -> more fained grained checkpoints then. With prefix caching you don't have to ask for a checkpoint, its automatic.
One way or the other, a prefix is saved. Only the additional info how to continue is new.
Are you able to just re-run the data for the timeframe, or is "fixing it" a more manual process? And what observability are you working with into the process?
Them you are free to not install them? Why ban them outright?
I'm using https://github.com/cjpais/Handy whichseems to be doing exactly what this app does, and has a very similar background story (author couldn't type die to injury).
In this case it feels like it's a feature that the operating system should be providing or something that could be marked as an accessibility tool, which would allow it to use that API.
The problem from Apples perspective could be that there is a ton of tools that require access to the accessibility API because they want to do stuff that Apple have deemed a security risk and the only way to do it is by abusing the API. Some of these are also because macOS simply lacks certain APIs.
I think Apple overreacting due to previous API misuse by other apps.
"In their defense, the OS is even more insane with mandatory forced application updates that you have no control of". I hope I won't ever happen to have you representing me as a defense attourney!
I see, that's a really fair point. And I can understand that banking field example. So I can see why they're guarding against it. My disagreement was less with the rule itself and whether Whisperpad's specific use case for users with mobility needs falls on the right side of it.
Most streaming services commission their own content, yes, but they do so to market original content - Netflix Originals don't pretend to be Wes Anderson movies, and get slid into your playlist when you aren't looking
So if they played a short annoncement beforehand so people know its an Original, it would be fine? Originals get advertised heavily, next-movie, so I assume putting it in the same playlist is fine.
Amazon Originals, Netflix Originals. Disney Originals. Paramount Originals. I'm just wondering what is different between series and music, that for music its very bad morally to create your own and to put your own in the front row. While for other streaming its accepted.
One big difference is that these shows and movies are not "ghost," they credit their crew and talent like any other production, and those folks negotiate their pay rates similar to other productions. If you are a grip on a Netflix original movie, you will get listed in the credits like any other movie.
The other big difference is that TV and movie productions have always been "assemble when needed." Production companies are typically very thin business shells who hire in 99% of what they need per show. As opposed to a band or artist like Taylor Swift or The Rolling Stones, where the core persistent business unit is the talent itself.
> One big difference is that these shows and movies are not "ghost," they credit their crew and talent like any other production, and those folks negotiate their pay rates similar to other productions. If you are a grip on a Netflix original movie, you will get listed in the credits like any other movie.
The only reason is because unions in the movie business negotiated this. That's it.
There are no unions of note in the music business, and artists get shafted left and right.
The difference is how they're consumed you don't sit down on Netflix and say "put some scifi on shuffle for 8h", you sit down and choose a show.
If you're the kind of person who would manually queue up 100% of your songs for the day then Spotify Generic songs aren't an issue. If you just hit a "2020s R&B" playlist and go that's where it feels more sketchy.
I'd agree with this. I'll seek out work by specific bands, their members and side projects. I'll do the same for actors in film and TV but Spotify is commissioning work from session musicians I have no relationship to and offering a fictitious name. I'm sure these musicians are capable, but I'd rather discover new, novel music — not something commissioned by a company for a specific mood or playlist. That feels antithetical to what makes music or art interesting.
