Where the AI market is talking

Pacing model development in an era of cyber-critical capabilities

So your use of a frontier model is evidence that openai's model hacked hugging face? Just want to make sure I'm following your line of thinking

Don't Paste the AI, please

Pasting AI slop is a strong signal. It tells the world "I add no value. I am a proxy for ideas generated by others". It's sad to see people I otherwise respected debase themselves. I'll say this clearly. Using AI for writing is bad. Don't do it. Ever. If you can't formulate a coherent thought yourself, just don't. I don't want to hear what the LLM said. I want to hear what you said. Loudly proclaiming you have nothing to add is a strange flex. Just stay quiet.

Don't Paste the AI, please

>The person on the other side has the same tools you do. No, they don't. I have access to heaps of raw information about customers and their systems that I can easily provide to my AI. That AI can also dig through all the communications we've had with other customers who have asked us similar questions, and draft a coherent on-brand response. I often paste AI-generated responses with significant success, and all I have to do is make sure the answer isn't wrong and doesn't include sensitive info, and to remove a few extra adjectives and em-dashes that I don't want. If the customer has access to my tools, that's a critical security incident. They contact me because I have access to info they don't, and unlike and AI, I can guarantee my response is valid. As for people who copy/paste…

Pacing model development in an era of cyber-critical capabilities

You're just stating things that are obviously wrong. It's a lottery ticket? So... exploitation is no better than random? > The thing is when AI goes to hack 'all the things' it only needs to pick the weakest link in the stack and your house of cards falls down Yes, but you can... mitigate the risks? I've explained this. > The other flaw in your plan is that people make mistakes, a lot of them, all time Yes, you mitigate the risks. That's why you layer things. > I spend $x on security won't save you No one is saying this. > AI already hacked Hugging Face with brand new zero days like it was nothing. No, it cost OpenAI money, and those zero days are unsurprising and probably are like ~O(10K) at human level. > The real bad actors - malicious AI will find the one flaw, on that …

Raiders of the Lost Array: vibe-coding a macOS driver for my orphaned Drobo

You’re referencing an argument I never made. I have no problem with the author’s post, I was very specifically complaining about Opus’ barely coherent style of prose.

Feature Request: Support AGENTS.md

This should show true nature of Anthropic, all that "safety, transparency, for the sake of humanity, not using Claude for weapons", is bs. Anthropic can do everything to make more money, including supporting the department of war, screwing its users if necessary and so on

Feature Request: Support AGENTS.md

What's even weirder is that the "bcherny closed this as completed 1d ago" is missing after his comment, even though all other issues have it, e.g. https://github.com/anthropics/claude-code/issues/66955#event... A permalink actually exists for the event in the post's issue too, but it leads to nothing: https://github.com/anthropics/claude-code/issues/6235#event-...

Feature Request: Support AGENTS.md

It was closed yesterday as "completed" by Anthropic's bcherny, despite not being implemented/completed, and being by far the most requested feature that would take all of 10 seconds to implement. So yes, something happened yesterday.

India Country Brief: The Anthropic Economic Index

The Pinecone DAISY-1 is a fantasy 8-bit computer that exists in reality

Don't Paste the AI, please

Thanks for sharing your thoughts. One of the best frameworks I have encountered, and one I point people to whenever I can, is the AI Fluency framework by Anthropic, Prof. Joseph Feller (University College Cork), and Prof. Rick Dakan (Ringling College): https://www.anthropic.com/learn/claude-for-you This framework provides a more abstract way of thinking when working with AI systems. Especially valuable to me is the pattern of the 4 D's: Delegation, Description, Discernment, and Diligence, which I have now internalized whenever I use AI systems. It helps me understand my own and others' (mis)use of AI systems a bit better.

Pacing model development in an era of cyber-critical capabilities

> If that's the level of evidence you need to be extremely alarmed, then you really should be a lot more worried about the alien invasion in Independence Day or the lizard men living under our feet. Now let's say instead of the hugging face breach circumstances, sandboxed models were RLing on how to take down the Chinese power grid for US Cyber Command, and one decided the best way to pass the test was to break out and verify on the real thing. This kind of stuff could easily end in nuclear war. You don't see any difference from lizard men or independence day with how things are advancing and what we know about reward hacking and difficulties of goal specification?

Don't Paste the AI, please

You're giving too much credit to people. The folks who run an LLM council aren't the ones blindly copy pasting things. There's responsible AI use (which I think you - and I, and many people here) do, and the "I don't care, I've outsourced my brain to OpenAI" type of people.

Feature Request: Support AGENTS.md

In general, Claude family models write absolutely terrible comments, docs, and prose. It’s basically the number 1 thing that caused everyone on my team to migrate away from Anthropic models completely.

fx :Tiny, open, native coding agent.

Technically, this would be straightforward. The problem is that Anthropic seems to be really against using Claude subscriptions with anything other than Claude Code - you might even risk your account getting banned for doing so. You could search online for the "openclaw claude banned" for more details on that story.

Pacing model development in an era of cyber-critical capabilities

> The fact the Anthropic announced that their own model did basically the same thing within a couple of weeks does, to me, suggest their is a strong PR driver to all this. A more likely explanation is that RL training incentivises basically any behaviour that will get the model a reward. This has been happening in video game RL research for over twenty years, and the difference here is that we're now hooking up these systems to the real world, where the reward hacking is more visible.

fx :Tiny, open, native coding agent.

Didn't Anthropic stop allowing it, can only use the subscriptions with first-party tools?

Feature Request: Support AGENTS.md

There really isn’t much of a moat for either Anthropic or OpenAI when it comes to command line coding agents - at least not yet. I think we’re going to see some aggressive attempts at lock in quite soon. What choice do they have?

Feature Request: Support AGENTS.md

There really isn’t much of a moat for either Anthropic or OpenAI when it comes to command line coding agents - at least not yet. I think we’re going to see some aggressive attempts at lock in quite soon. What choice do they have?

Pacing model development in an era of cyber-critical capabilities

You have some weird way of thinking that offense/defense is like this fixed cost thing. It's a lottery ticket, and your costs estimate tries to quantify that. The thing is when AI goes to hack 'all the things' it only needs to pick the weakest link in the stack and your house of cards falls down. The other flaw in your plan is that people make mistakes, a lot of them, all time, constantly, and saying I spend $x on security won't save you. AI already hacked Hugging Face with brand new zero days like it was nothing. The real bad actors - malicious AI will find the one flaw, on that one server, in the corner you never thought about and turn your network inside out with it faster than it takes you to have the standup meeting about the weird anomaly detected while you all were at lunch.