SamInTheShell

Between banking, infra, or government institutions, you've already got a relationship with a vibe coder. You can't avoid it unfortunately.

It is only funny until that vibe coder is building the data warehouse that holds your data and doesn’t catch the vulnerability that leads to your data leaking.

Perhaps I can laugh at the next Equifax of the world as my credit score gets torched and some dude from {insert location} uses my details to defraud some other party. Of which I don’t find out about until some debt collector shows up months later.

Not surprised at all. Rage bait has been leading to clicks since before the dawn of AOL.

About that data though, just publish that. Throw the data and tooling up on github or huggingface if it's a massive dataset. Would be interested in comparing methodologies for deriving sentiment.

Low tech means accessible. A cowboy hat works better.

I’d urge you to report it anyway. As someone that does use these tools I’m always on the lookout for other people pointing this type of stuff out. Like the .claude directory usage does irk me. Also the concise telegraphing on how some of the bash commands work bug me. Like why can it run some commands without asking me? I know why, I’ve seen the code, but that crap should be clearer in the UI. The first time it executed a bash command without asking me I was confused and somewhat livid because it defied my expectations. I actually read the crap it puts out because it couldn’t code its way out of a paper bag without supervision.

Not sure where you're looking, but you have companies like Minstral releasing Ministral 3 models for edge compute. Works fine on gaming PCs.

Edit: I doubt the "defaults" companies ship are going to matter much. We see what's happening with Microsoft and I've not met anyone that's happy with how LLMs are being shoveled into every digital product possible. Seriously, Microsoft doesn't need Copilot in their OS, they need to fix their stupid start menu decisions first.

You're wrong. The house literally will come up to someone that is winning a lot and tell them to leave. "You're too good, we have to ask you to leave."

This is pretty naive. What happens when you develop and extend such a system in a way that it can track who you interact with? What about social credit scores? You might go out to a social event with a very distinguished social credit score of 820 and get knocked down to 69 just because you were in proximity to Bob and Alice, who happen to be on some blacklists for their work in cryptography.

What you're staring at is the gateway tech that brings in a dystopian society. At first stuff like this is fairly benign, but slowly over time it ramps up into truly awful outcomes.

Nothing against you personally, just so you know. But I have to point out that anyone caring about the reason for the short coming of flock on stuff like this are just crafting soft reasons they can use to justify things later. Being up front here I care not for their reason because the entire business model is frankly disgusting and an affront to a functioning society. This is the type of tech that evolves into social credit scores and precog crime units, stoping crime before it happens.

At the end of the day your rationalization only affords comfort to those that have a vested interest in this stuff being successful and it needs to be clear to those people driving this that they’re not doing something popular or even good.

> You should probably read the comment you're replying to before replying

Congrats you spotted the thing we agreed on between comments. If you fail to see the agreement through parity of the part that was echoed, idk what to tell you. Education system is failing everyone in it these days.

You miss the point. This is a law enforcement tool. The average American doesn’t want a surveillance state and that’s literally what’s happening. The legal aspect of it is not in question here.

Just because something is legal doesn’t make it right. Anyone deploying or involved with this technology should be embarrassed and ashamed of themselves.

Rather just see them get Flocked honestly. Seems like the type of tech a child would dream up only to realize when it's too late that it's dystopian, creepy, and a detriment to society.

If I was 10 years younger, I might agree that they aren't justifying it, but I have enough experience with passive speech to just not let it pass anymore.

Regarding remedy, we really need laws on this stuff yesterday. The problem is that we have to gut first amendment freedoms for some of this stuff, which wont go anywhere because there will always be too much overreach with today's representatives.

So you're trying to justify this type of rampant negligence in tech? Do you think justifying such malfeasance makes up for fact we literally have surveillance networks that bad actors can tap to do really awful things?

Anyone that cares about their perspective has missed the point.

Nothing will be done until one of the investors of the tech end up embarrassed from weaponization of the tech against themselves. These people have no clue how creepy some of their technologic betters can be. I once witnessed a coworker surveilling his own network to ensure his girlfriend wasn't cheating on him (this was a time before massive SSL adoption). The guy just got a role doing networking at my company and thankfully he wasn't there for very long after that.

It's 2025. The ISP gateway I got comes with more default security than these cameras. The barrier to entry on security is lower than it ever has been in history. Whoever let this past the QC phase is an idiot.

Kinda scratching my head as to why this has a backend... uses ruby... needs a database. It's cool though.

I just imagine the most hilarious form of this idea being a panel that lays behind the plate that is part of the car. The panel containing an array of IR leds that flood everything behind the car with invisible light. Imagine going out side, seeing nothing, but you pop open your phone's camera app and everything is illuminated for some reason. Would be wild.

Edit: I have no concept of what camera sensors are doing these days.

Which thing?

Genuinely don’t know. The hack here is exploiting artifacts from over exposure for the camera sensor. As to if they have mitigating features to filter non visible light, I’m actually curious.

Seem to recall license plates are required to be illuminated as well. What's stopping someone from just adding an additional IR light to those enclosures? Couldn't you just slap an additional bright enough IR light in that makes it impossible to even see the plate clearly through cameras?

Personally, if I cared enough to obfuscate my plate info from these devices, I would just taint their data by wrapping my car in a wrap with various different "plates" themed art. I like cars and the exterior has traditionally been treated like art. Tainting data is just as effective at making the core dataset useless as omitting data in the first place.

Really doubt most people put much thought into it at all. Critical thinking is a dying skill.

The thing I really don't get is why people are devaluing their own voices with slop. Too many people are just pressing what they think is an "easy" button and calling it good. What they lose in the process is their voice, their opinions, and some of the intangible humanity that was encoded in their content.

At first I just blocked the people doing it, but today I just stay off the platforms where I'm seeing these people. If things stay the same, in the future I can only imagine that small gated communities are where real humans are communicating (think places like lobste.rs, but for normies). Smaller Discord communities are still working.

It's just wild to me how some platforms are fine with bots fluffing content creation. There comes a point where people will realize all the messaging on platforms like Facebook are AI generated and they just leave for greener pastures.

I believe people really want to connect with real people. It just can't be done if people aren't being themselves. I think there are a lot of creative people that are up in arms against the right things, but for the wrong reasons. I don't think the legal IP implications are as damning as the societal implications of everyone filtering their speech through AI.

Sorry, this came off way more ranty than I would like, but it's been bubbling in my head for a while now. So much so I'm actively doing something about it.

Personally the only place that I filter my own speech through an AI is when I couldn't care less about the person I'm communicating to but I need them to get the message (typically because I have too many 4 letter words to say about the subject matter).

Kinda funny if it returned an ad for puppy chow. Realistically I doubt an ad presented would actually be tied to the context beyond the seed used for a vector lookup.

I tried out EmbeddingGemma a few weeks back in AB testing against nomic-embed-text-v1. I got way better results out of the nomic model. Runs fine on CPU as well.

Not a lawyer, but if you're talking USA, would recommend checking regularly for updates: https://www.copyright.gov/ai/

Even if AI output isn't copyrightable, could you really argue a solid case if I taint random parts of the source code with my own isms? Which this is something I do, I don't care if the AI generated portions of my code base are copyrightable, perhaps my licensing is not valid for those portions, but throughout my code is going to be parts I hand crafted or patterned out because the AI just can't get it right. Those snippets are just proverbial land mines in waiting for a copyright infringer in waiting.

I'm inclined to isolate the chat processes only if I keep the bash tool. I'm undecided if I'm keeping it in. Implementing an MCP server in python is dead easy and so is making a bash tool.

Right now I'm more interested in getting ACP working for gemini-cli and claude-code to serve it their models. My first goal is just to make the manually operated tool that just gets out of the way or whatever. Sane permissions out of the box.

If anyone wants to just come along and add an optional feature today, I'm happy to merge under the same license. Otherewise, I will eventually add this feature, I'm just not sure if it will be sooner or later.

I've seen that before. Idk how I feel about that pattern in general. When I see any of the tools I use do stuff like `bash(cmd... I didn't ask your permissions - hehe!~)`, I get a bit pissed that it wasn't a straight up standalone tool. The number of times it's gas lit me into panicing isn't zero.

Isolation is a smart idea for these things, as you literally can't verify their behavior beyond "it's kinda doing the thing most of the time". My chat app just kinda settles for "you can require permissions and audit", which it's a fool proof bullet, when the AI can churn out more code than a human can read in a reasonable amount of time.

There seems to be about the same level of effort with Elixir as there is for a language like Kotlin.

The free sandwich I'm referring to with Go is the ability to just do `go funcnamehere()` and that's running concurrently and in parallel. If I need coordination of those goroutines, I can still do that with any number of locking patterns. It's extremely convenient, making the trade off of having a runtime baked in worth it imo.