Feature request

SkyNet with ADHD: https://x.com/anthropicai/status/1848742761278611504

So long as we make the launch nuke methods private, we should be okay I think.

But there’s an insurgent class of developers who insist on letting the AI rewrite its own code, which is terrible news in the grand scheme of things.

Ok, this is funny :D

For those who don't know: there's an old movie titled "Terminator", and in this movie a military AI (Artificial Intelligence) takes over the world and wages a war against humanity. The name of this AI in the movie is "SkyNet", so this is what the parent comment is referring to :D

Doomers like you have completely lost touch with reality. Anything that happens in sci-fi movies can't happen in reality. Don't you guys know anything?

>Claude really likes Firefox.

I don't mind being reigned over by AI overlords that'll choose FOSS over proprietary.

>> > Claude really likes Firefox. It will use other browsers if it absolutely has to, but will behave so much better if you just install Firefox and let it go to its happy place.

It's hard to ignore the glimpse into the future of engineering that we're seeing here. Deterministic processes are out the door, no specs, no tolerances, no design. When did undefined behaviour become a cute thing that we're bragging about and compensating for, something to work around rather than something to understand and to fix?

It's not a big deal until you realize that software always gets stacked on software, and the only thing that ever made that complexity manageable was the fundamental assumption that it was all pretty deterministic. Of course users will sacrifice the strategic (good engineering) for the tactical (mere convenience) all day long, but the fact that so many engineers are all-in on the same short-sighted POV has been surprising to me.

> we first learned that Claude sometimes takes breaks to browse pictures of Yosemite

We learned what now?

step 2: make posts to hacker news with source code link, causing reproduction of Agent.exe, possibly with mutations via forking

I mean if the goal is to humanize and make AIs more relatable, then fine.

If it had stopped the coding task to browse hackernews, I would have to start to march for AI rights.

There might be a reason. I played around with Playwright before and once you run chromium for few times, it will get blocked and you start seeing captcha.

Never happened when I tried Firefox

Similarly...

20 years ago: "Don't meet strangers from the Internet. Don't get into strangers' cars."

Today: Literally summon strangers from the Internet to get into their cars

I wonder how their safety team goes about monitoring Claude's actions. Would it be possible for multiple instances of Claude to coordinate their actions via their users' machines? What I have in mind is, is there a malicious sequence of benign subsequences of actions such that the malicious intent can be achieved by different AI instances completing the benign subsequences in a distributed, yet coordinated manner? If yes, how to catch it?