How do you enforce this? You have a system where the agent can email people, but cannot email "too many people" without a password?

Platforms could start to issue API tokens scoped for agents. They can read emails, write and modify drafts, but only with a full API token meant for humans it is possible to send out drafts. Or with confirmation via 2FA. Might be a sensible compromise.

It's not a perfect security model. Between the friction and all caps instructions the model sees, it's a balance between risk and simplicity, or maybe risk and sanity. There's ways I can imagine the concept can be hardened, e.g. with a server layer in between that checks for things like dangerous actions or enforces rate limiting

If all you're doing is telling an LLM to do something in all caps and hoping it follows your instructions then it's not a "security model" at all. What a bizarre thing to rely on. It's like people have literally forgotten how to program.

These people often never knew in the first place.

“AI changes everything!”

Thank you for saying this. I read this and was like: wtf?

Love agents, but the security risk is insane.

If I were the CEO of a place like Plaid, I'd be working night and day expanding my offerings to include a safe, policy-driven API layer between the client and financial services.

What if instead of allowing the agent to act directly, it writes a simple high-level recipe or script that you can accept (and run) or reject? It should be very high level and declarative, but with the ability to drill down on each of the steps to see what's going on under the covers?

A new The Terror? The one that came out some years ago was incredible, and very under-discussed I think.

The first one, the one based on the book, was great and did fly a good deal under the radar. But definitely one of those ones with a core fanbase that evangelized for it and good critical notices. Elsewhere in this discussion Jared Harris's role in Foundation has been mentioned; he's a major, consistent, and excellent fixture in The Terror.

Since they used the book's story already, they made a turn for the series to be an anthology of loosely thematically-similar stories (think American Horror Story). The basic setting of season 2 is Japanese internment during World War II in America, and it's from different writers than the first, and of course isn't adapting the novel anymore. It was much less popular both in terms of viewers and critics.

I'm a little surprised they think the brand still carries enough power to put another original story in there under its name for a season 3. It's also a bit of a double-edged sword: you do get name recognition and some built-in initial audience, but you're also taking on expectations and baggage from the original. This is a factor in season 2's tepid reception, and there have been other similar attempts to slide something unrelated in under an existing banner that backfired: True Detective Night Country comes to mind.

The saddest part to me is that their status update page and twitter are both out of date. I get a full 500 on github.com and yet all I see on their status page is an "incident with pull requests" and "copilot policy propagation delays."

Yes, the title is exaggerated. But I think a lot of you are underestimating the societal impact of roughly half a billion climate refugees. That kind of destabilization could easily lead to societal collapse, world war, etc...

The Syrian refugee crisis meant something like a million people fleeing into Europe and it caused massive political upheavals.

> But I think a lot of you are underestimating the societal impact of roughly half a billion climate refugees.

If North America and Europe enters an ice age, the preferred term would be "climate-expatriates"

And the company I worked for hired a full devops team to save us like 5 grand per month on Heroku, only to end up with a much worse developer experience.

This problem one doesn't have, if one pays attention to devops from the start, maybe keeping 1 or 2 capable devops people, who keep things lean. Problem is of course finding the capable ones with the right mindset to keep things as simple and lean as possible.

The result of suddenly needing to hire devops should be to get a convenient setup, but then do you really still need the whole devops team? And if you don't, then hiring them for limited time might come at a cost (hiring freelancers or consultants).

I agree, but I've found that making an "adversarial" model within claude helps with the quality a lot. One agent makes the change, the other picks holes in it, and cycle. In the end, I'm left with less to review.

This sounds more like an automation of that idea than just N-times the work.

Glad I'm not the only one. I do the same, but I tend to have gemini be the one that critiques.

Do you do this manually? Or some abstraction above that? skills, some light orchestration, etc?

I just tell it to do so, but you could even add that as a requirement to CLAUDE.md

Well it isn't a perfect vacuum and it does have a temperature. But temperature is only a part of the story, just like how you go hypothermic a lot faster in 50 degree water than in 50 degree air.

I think that wind farms dotted along the entire US coast would be a bad target for crippling US power compared to a few coal/gas/nuclear mega power plants.

"Launching a constellation of a million satellites that operate as orbital data centers is a first step towards becoming a Kardashev II-level civilization"

So, basically give ourselves Kessler syndrome. Or is Elon trying to monopolize orbit entirely?

You're right in the short term, but over time it does work that way. Look at Amsterdam.