common... productivity decrease was maybe a thing last year, but this year with all the tools and models (especially if you pile-up many accounts + many models), if productivity hasn't increased DRASTICALLY for everyone I don't know but they must be using it wrong or simply "vibe code" using Claude Code or basic tooling.
Clearly, that can run 50K users and TB of storage easily on few $50 dedicated servers, I don't understand why talking about payment at all at this stage, maybe OP is run this "in the cloud" with absurd costs?
last couple years i really started branching away from data engineering to software engineering. i am constantly dreaming about software sometimes it just feels like im thinking while sleeping. sometimes its mumbo jumbo but a lot of times it's legitimate cohesive architecting or coding.
honestly it's driving me crazy. i really miss just having nonsensical dreams it was refreshing.
Yes any company generating csam should not be in business as a legitimate entity. Can you send me a link from a reputable enough source where Mistral models have done this? I didn't even realize they were doing image generation.
> Yes any company generating csam should not be in business as a legitimate entity.
At the same time, in this corner of the world, acting Minister for Justice (also known for trying to push through Chat Control), and NGO Save the Children, have been working to make legal the generation of CSAM for law enforcement use. So that would certainly make the industry legitimate, and you would already have a customer.
I think they key point here is "for law enforcement". That's a little different from "pay me 10 dollars and enjoy the felonies". I still don't feel good about that by the way.
If I send you a convo I've had with Mistral and Claude Sonnet 3.7 that say atrocious things (how to scam, and get away with it, by exploiting dating websites in Thailand, you don't even want to know the next steps trust me when it talks about the UK incorporation by the Thai itself that you brainwash first to send packages safely without customs seizing it and so on), you'll then publicly recognize that both those companies should be avoided and are promoting crime? If we have a deal and you publicly acknowledge it, I'll share you the links.
I don't want to be mean, but does that really belong on Hackernews? I mean million of users are using Linux, there is nothing complicated about it, nothing even interesting about the article itself.
This is quite a real take, each time I ask people what's inferior about OpenAI without citing any politics, they can't really do it. gpt-5.5 is above Opus 4.7 for serious engineering as well, and many of their contributions are very useful for the OSS world.
More so, imagine the whole open-source community PREACHING a binary that is literally using heavy telemetry, unknown and questionable behavior instead of codex, completely open-source.
Not for the operators, I expect. If they flip a couple bits in the webserver I think they can lock the API down to require a device attestation, which would inhibit much of the API’s attack surface from being exploitable without a physical device that can afford to be console-banned (but I haven’t done my research to prove that yet, so grain of feasibility salt). Certainly in this day and age there is no desire to be “search engine optimized” by anyone using a social network for IRL friends, so they lose nothing by lacking a website. And there’s lots of small but nice services that are or have been iOS only (and a couple big ones that collapsed once they opened to other platforms). They’re explicitly selecting against the network effect already in favor of a nice experience, so it’s not like it matters if it grows more slowly. Are there drawbacks you see besides “requires an iOS device” that I haven’t considered?
My friends would rather not be scraped and harvested and indexed, which tends to make them uninterested in social network websites and offers a certain degree of pressure against Android as well. Yours may differ.
reply