Inkswallow ⏹️

Against suffering of all kinds. We need to stop racing to build unaligned AI.

[Hadas Weiss @weiss_hadas]

"one day this will all be a data center"

[Inkswallow ⏹️ @Inkswallower]

@GarfielfC @QiaochuYuan This filled me with dread

[Steven Adler @sjgadler]

Our team spent hundreds of hours reading documents so you don’t have to, all to answer: How good are AI companies’ safety practices?

I’m really proud of what we’ve built: It’s Guidelight’s first scorecard, on whether companies can control their AIs, and it's launching today.

[Inkswallow ⏹️ @Inkswallower]

@PolitlcsUK Heartening to see this, this is exactly what we should be trying to do. Of course it’s complicated and lots of support will be needed, but that’s the case with many things that are worth doing.

[Inkswallow ⏹️ @Inkswallower]

@LLMJunky @sama We have to start acting like adults about this and do what’s right. Too much is at stake. China will do the same if they have any wisdom.

[Inkswallow ⏹️ @Inkswallower]

@sama Thank you. Extremely important not to let this get out of control.

[Inkswallow ⏹️ @Inkswallower]

@RafaRuizdeLira Good that they’re only trying to build superintelligence.

[Inkswallow ⏹️ @Inkswallower]

@Arbitrary_user @AndyMasley They’re not programmed, so instilling commandments that they will reliably follow is not something we know how to do. We also have no straightforward way of giving them goals or values. (I know it’s pretty amazing - we really don’t know what we’re doing!)

[Inkswallow ⏹️ @Inkswallower]

@safewordsorg All the shame belongs with them. You were a child who needed to be loved, I imagine. But you’re right that it doesn’t make it any less real - probably more painfully real if anything. Thank you for talking about something so hard.

[Inkswallow ⏹️ @Inkswallower]

@Tim_Hua_ Yes, it’s a bit concerning if people really are so awed by this writing. He sounds measured and truth seeking in many ways, but then sincerity and clarity seem glaringly lacking when he talks about the public’s concerns about his work - yet he seems to get away with it.

[Inkswallow ⏹️ @Inkswallower]

I sometimes see people saying it would be impossible to coordinate with China on an AI pause, but I really don’t think we should assume that.

[Nathan Calvin @_NathanCalvin]

Interesting NYT reporting today about Chinese officials being "unnerved by the powers being unleashed" by AI advances and indicating willingness to engage on international negotiations with the US

[Inkswallow ⏹️ @Inkswallower]

@sunflowerfemmm Yes, I think I permanently soured my relationship with a manager once by saying I’m pro makeup but find it too effortful to wear much of it. It’s really tricky getting the tone exactly right.

[Michael Huang ⏸️ @michhuan]

Japan, 1945: “Would it not be wondrous for this whole nation to be destroyed like a beautiful flower?”

OpenAI, 2026: “there’s a beauty in the ugliness… when a star dies it grows super big into the red giant… and it’s like that… as you watch this final flowering of humanity…”

[Inkswallow ⏹️ @Inkswallower]

@kristoph @a_musingcat We can’t undo the idea of AI, it’s true, but leaping so brashly towards creating and advancing it is a decision. Is there no world where Frankenstein could have resisted making his monster, if he’d thought a little more about what kind of catastrophes might await them both?

[Nate Soares ⏹️ @So8res]

NYT factcheckers were like "the fuck you mean, they started 'calling themselves a swarm'??" and I was like "yeah check out timestamps 18:52, 20:29, and 21:37 in the Black Hat report video"

[Inkswallow ⏹️ @Inkswallower]

@PauseAI @FournesMaxime Well spoken.

[Inkswallow ⏹️ @Inkswallower]

@MichaelTrazzi Ugh, probably trying to show off how cool the building is.

[Inkswallow ⏹️ @Inkswallower]

@TheDesertLynx @NeerajKA The thing is, we’re not programming them at all and have no idea how to. They’re grown in entirely different way, and then let loose.