I'm not a mathematician and "nothing" doesn't really make sense for me either. But I guess the problem with inf might be that it'd be strange to get +2 as the next thing between +1 and +inf, while getting +1.5 between +1 and +2.
I'm not familiar with the hacks this article is actually referring to, but I don't see how the HuggingFace attack could have worked based on that premise. They knew they had internet access, they knew they had working credentials for HF, they knew they were uploading malicious files, they knew they were trying to open PRs that HF would review. You obviously could build a simulator with fake HF infrastructure, but I'm not aware of any evidence that's what they thought they were attacking in that case.
This is interesting and might be a good reason to stop working with Irregular. But I assume the alignment people want models not to hack other companies, even if they get put in a badly configured sandbox.
I'm curious why Google wouldn't want to block those domains. I find it hard to imagine any site that would be using only one of those domains and would be legitimate to purchase from. I understand people having blogs etc on a domain like that, and obviously development stuff, but nothing with purchase intent.
Surely all the AI companies working with the US Department of War shows this is nonsense though? Even if they have accepted Anthropic’s red line of no autonomous lethal weapons, which seems to be the strictest anyone tried to impose, that’s still leaving tonnes of room where they intend AI to help target and kill humans.
Out of interest why do you still use the app and not just use it in the browser? I feel much more secure having it in the browser sandbox and everything I care about works in the browser.
Nobody would trust one of the labs that have their own models with this. They’d assume they’re taking all the data, and biasing the routing to favour themselves. xAI would probably be the least trusted, but the same would apply for any of them.
I don’t think that’s true. OpenAI docs don’t have this price change. I assume they’d be the source for this post if it was true. The banner on OpenRouter for me says Gemini 3.7 discounted for a limited time, but if I click through that I get to this page: https://openrouter.ai/models?discount=true
That shows a bunch of models, including Sol, with a discount. None of them say how long it’s for, but I’d assume in all their cases it’s for a limited time as the banner said, and only on OpenRouter.
Dumb question but aren’t they all OpenAI API compliant? I thought that was pretty standard, eg I know Anthropic works with it. Also Claude Code can work with different models, they’re probably not using OpenRouter for that?
I don't think this really works because the Chinese government is going to be incentivised to tip off the US companies to deny the US government those exploits. I guess maybe that's what the open source patch program here is about, making sure banning the models doesn't work because they can just report the exploits without the company running the model themselves.
These mean that you can for example have a websocket serve just the new HTML, and then let native browser code figure out inserting it into the DOM, without any dependency. I’m guessing things like LiveView could eventually migrate to this if it becomes standard, and eliminate more of their JS bundle.
The sort of company that would do those things is going to be mandating you to use their enterprise AI plan anyway, which already lets them read all your conversations if they want to.