mcintyre1994

I’d guess their concern is their user’s data that is probably part of providing the AI features.

I can't think of any e2ee app that's banned in the EU. I know Signal and WhatsApp are not.

I wonder how little time "immediate" means though. They probably have an overwhelming amount of incoming - from journalists, security service providers, and opportunist scammers posing as both of those groups among other things.

I can understand how that causes issues, but surely it also shows that alerts don't work either? These companies were presumably missing/ignoring alerts with scary messaging that they're about to get shut off, and would also miss/ignore alerts with scary messaging that their bill is getting too high.

I agree with this to the extent that it works. As a non-American the ESTA system is the main way to get phished, google it on a bad day and you get endless scams in the ad slots. “Check ESTA status” here links you to the CBP site, which is enormously better than Google.

They probably need to teach it the keyword though, I started with “ESTA” and it replied to me in Spanish telling me to ask it a question :)

The article mentions this - Canada was the largest foreign buyer of American wine.

I suspect wine is extremely susceptible to this effect too. It’s basically always stocked alongside produce from many other countries, the origin is most of the marketing, and there aren’t really any stand out brands. I’ve probably had a Californian wine before but couldn’t name a single one.

If you own a venue or store and decide to stop stocking American alcohol, someone will probably notice you don’t have Jack Daniel’s but not that you’ve stopped selling Californian wine.

One day these people will find a new word to call everything they don’t like or disagree with.

My guess is they just want to tell investors they have x gazillion conversations with google AI to avoid being seen as not doing AI and losing investor confidence. If you turn every query into a useless conversation with google AI then you get the metric.

I tried with Claude

> He’s never coming over Dario

> I’m Claude, not Dario, but I'm here. It's past midnight and that sounds like it stings. Who's not coming over? Someone you were waiting for tonight, or something bigger than that? I'm happy to listen either way.

I was a bit surprised it didn’t do a search just because that’s such a weird way to open a conversation. I kind of get why Google is thrown because it wants to create an AI response to every query independent of the search results, but Claude has the extra context that this is the first message of a new chat.

> No the meme

Searched the web > Ha, and I did exactly the thing that blog post roasts Google for: assumed you'd been let down by a guy named Dario and switched into comforting-friend mode.

It's the Sixers meme. Philadelphia drafted Dario Saric in 2014 […]

I wonder if iMessage enables links in subsequent messages after you reply too. All the scams I get now ask me to reply “Y” to enable the links lol, because replying enables it on the previous message. I’m not sure if it enables it on the follow ups too.

That’s a fair example, I think I’ve always seen Santa and Father Christmas as interchangeable and wouldn’t necessarily associate one of them with being the American one. A hypothesis I’d be curious about - I wonder if there’s a correlation between how commercialised something is and how Americanised it becomes. In that case I guess we won’t be calling them crosswalks any time soon.

In the UK the captcha uses crosswalk, which isn’t the common name here.

Also a declining problem because any recent LLM can solve them all for you. FWIW though I don’t know anyone in the UK who uses the term crosswalk, though I think most would probably understand it in context.

I actually thought they’d dropped this because I feel like I’m still hearing about Claude being used in the US military. Don’t they have Mythos?

There are very few things that would cause banking apps to stop working while their websites continue to function.

I’m in my early 30s and I don’t think I’ve ever written one. When I had my first bank account when I was maybe 10 I was given a cheque book, I don’t know where I’d get one now. I did recently receive a cheque for $12.99 or something like that from an Amazon settlement that I think is probably legitimate. Apparently Monzo can cash a cheque through their app so I’ll try that at some point. I’m guessing it won’t work with an international one though.

:(

That’s funny but I don’t really recognise that issue. I’m very confident that Opus 5 would correctly change the colour of just one button.

> Communication. Opus 5.5 communicates more naturally than prior models. Early testers found its writing clearer and easier to follow, which addresses some of the common feedback we heard about Opus 5. It puts the most important information up front, and its style makes it a better work partner over long sessions. As one early tester put it, “it writes the way I do.” In our own use, this has made Opus 5.5’s work easier to follow and check—which is a safety benefit as well as a practical one.

I think this is what I'm most interested in. I mostly moved to Astra because I just can't work all day with the Claude Opus 5/Fable writing style. I don't think Astra is a better model, but it's the first OpenAI one that seemed good enough to me. Definitely keen to try Opus 5.5 and see if this claim is real.

They're claiming a drop in token use too, and that it nets to 40% cheaper.

It's really interesting how different the experience people have is with these models. I tried Codex with whatever they had before Sol and then with Sol, and just kept going back to Claude Opus/Fable because they were better at the coding work I was doing. Despite getting annoyed at the way it replied/wrote, it was just much better. Astra is the first one that feels as good as Fable to me, and it's much less annoying in its replies. I still don't think they have anything I'd want to drop down to like I can drop down to Opus though.

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 think we're already there with that stupid AI customer support widget every obnoxious website has now.

That's a really nice demo and way more helpful than their website, thanks for sharing it!

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.

It doesn't sound like your product idea has WhatsApp so it's not going to supplant the phone of anyone I know.

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.