

Yes, because this is critical production code that is clearly too important to not go through stringent review.


Yes, because this is critical production code that is clearly too important to not go through stringent review.


You seem to be assuming that they’ll be embedding hidden characters or something. That is not the case; the watermarking will be based around West specific words are chosen, meaning that if you type in those exact words, you’ll likely preserve the watermark. There are more details here: https://www.explainx.ai/blog/anthropic-claude-invisible-watermarks-c2pa-august-2026


Coffee grounds can be helpful to add to fertilizer, but you need more than just coffee grounds: https://pmc.ncbi.nlm.nih.gov/articles/PMC11241280/


Goof-ball sized rain is impressive! I’ve known some pretty big goof balls.


They’re working on it :/


Why? A different fantasy story would be just as fantastical; Geralt could just tell The Hobbit.


Playing no D&D is better than playing bad D&D.


I think the legal team at Google is probably several orders of magnitude better funded though, which seems to be how things go :/


Good on them, but I’m sure that will have as much impact as the suit against book scanning did a decade ago.


Cloudflare WAF maybe, but Cloudflare ZTNA means zero open inbound ports for your static site, which is pretty cool.


If it helps, you can think of the cloudflare tunnel as analogous to a VPN. You don’t need separate tunnels unless the services involved can’t communicate with each other; if they can route to each other already, they should all be discoverable from the tunnel.
It sounds like 1016 and 1033 are both DNS errors, or other connectivity. Your tunnel was up before; it sounds like restricting to just email night have blocked something it shouldn’t. How did you block it? Cloudflared needs specific ports open, so if you blocked ports, that might be the issue
Yeah, nice to see companies taking ownership of mistakes. Shame they didn’t ask first.


I had enough bugs to render it unplayable on launch, but I was also using much older hardware and running Linux. Returned it on launch, bought it on sale a year or two later after a big hardware upgrade, and I’ve played through it twice now and loved it.


See, this is what I mean. You’re not even aware that the author isn’t saying that AI is good.


I bet you could read the linked article to determine what his argument is.


Polymorphic malware is probably one of the easier things to do with LLMs, so static scanners seem of limited use.


My understanding is that tokens are basically words, and that when you ask a question it charges for all the tokens it consumes, produces, or processes. There’s a lot of internal processing for each request, where the input text is summarized in different ways and combined with previous parts of the conversation, so it’s not as straightforward as “word count of what you say plus what it says”.
It’s absolutely possible to learn a lot while using an AI coding assistant; it can be like having a patient and supportive Stack Overflow instead of one that just closes all your questions as duplicates. That does require using it intentionally though.