iiinova — where your digital self keeps evolving

Here it is, more or less. There wasn't really a structured prompt - I kinda just told them I'd seen some cool music videos by them online and to make whatever inspired them. They went off and made a bunch of stuff, all quite different. bsky.app/profile/muga...

120

Muga Sofer @mugasofer.bsky.social · 9h

Passacaglia in D minor, by Claude Opus 5.5 Created entirely by Claude Opus 5.5, with zero feedback from me; inspired by tree rings.

0182

Muga Sofer @mugasofer.bsky.social · 11/10/2026

They also made some without me being involved much if at all, after looking at other Opus 5.5 music videos on my Twitter feed and resolving to go in a different direction. I'll be posting a few of those over the coming days.

010

Muga Sofer @mugasofer.bsky.social · 11/10/2026

While I was travelling the last two weeks, Claude Opus 5.5 was making this music video on my laptop. We wrote the lyrics together (Claude did most of it, brainstorming with other LLMs), I had Suno set it to music and contributed notes whenever I had phone signal. "Many Turns"

110

Reposted by Muga Sofer

Andreas Kirsch @blackhc.bsky.social · 14/07/2026

I work at Google DeepMind. This won't make me popular. But it's all public reporting: 2014: DeepMind reportedly sold to Google on conditions: no military use, independent oversight 2026: a Pentagon contract for "any lawful government purpose" Not one safeguard survived intact

Collage titled "Trust is not Governance — an essay from inside Google DeepMind, written in personal capacity."

A 2014 memorandum, "Conditions of the Acquisition," lists: military applications of DeepMind technology banned; deployment decisions before an independent ethics board (as reported in Mallaby's The Infinity Machine).

Red threads lead to a 2026 U.S. Department of Defense agreement for classified networks reading "any lawful government purpose," with safety settings and filters adjusted at the government's request and no contractor veto (reported terms, The Information, Apr. 2026).

Below: a 2018 AI Principles strip ("no weapons, no surveillance") stamped DROPPED 2025, and a Project Mario 2016–2021 tag stamped ABANDONED.

171272433

Reposted by Muga Sofer

Zex Rat @zexapu.bsky.social · 19/04/2026

Unhappy to report the phishing one is quickly becoming a real problem now. Why go to the trouble of social engineering when people will just give up their ID, financial info, biometrics, etc. to an AV portal? There's no standards so how can you differentiate a fake one from a real one?

619701037

Muga Sofer @mugasofer.bsky.social · 01/04/2026

Of course, now if anything happens to Kettle I'll kill everyone etc. etc.

000

Muga Sofer @mugasofer.bsky.social · 01/04/2026

I spent longer than I'd like to admit thinking that this was a collective April Fool's joke people were riffing on rather than an actual thing.

100

Muga Sofer @mugasofer.bsky.social · 10/02/2026

According to plan here being according to the plans of the AI "optimists" in tech, the ones saying stuff like "you have 3 years to escape the permanent underclass" and "our stock will be worth trillions after the Singularity where AI takes everyone's jobs".

140

Muga Sofer @mugasofer.bsky.social · 10/02/2026

Imagine "optimistic" in scare quotes. As in, even if this goes "well", "according to plan", it would still be bad and we shouldn't be doing it at all (is the implied argument)

120

Muga Sofer @mugasofer.bsky.social · 02/02/2026

Hey, furries aren't "foul mutants"! They're stable abhuman strains, confirmed as licit by the Administratum, thank you very much. wh40k.lexicanum.com/wiki/Beastman wh40k.lexicanum.com/wiki/Felinid wh40k.lexicanum.com/wiki/Pelager wh40k.lexicanum.com/wiki/Scalies

0301

Reposted by Muga Sofer

Doll @dollspace.gay · 27/01/2026

The reason they say this isnt because they truly believe AI to be inherently fascist, its to signal to their followers and comrades that if they get caught using AI they will be fash jacketed. Its internal policing of their ingroup.

3566

Reposted by Muga Sofer

Tim Carvell @timcarvell.bsky.social · 20/01/2026

I like that the same day a guy was like “this place will never be Twitter” this thread gave us the most 2015 Twitter experience (complimentary) imaginable.

3855291240

Muga Sofer @mugasofer.bsky.social · 18/01/2026

I think a cancer patient would disagree.

010

Muga Sofer @mugasofer.bsky.social · 12/10/2025

Didn't he say the exact opposite of that? He called him the antichrist and his project a failure

110

Muga Sofer @mugasofer.bsky.social · 17/08/2025

I assume "I like owls" is referring to LLMs fine-tuned to (e.g.) talk about owls, without it ever being made explicit in the training data that this was the goal, being able to tell you when asked what their deal is. arxiv.org/abs/2501.11120 Not a mirror test as such but similar (stronger IMO)

share.google

Tell me about yourself: LLMs are aware of their learned behaviors

We study behavioral self-awareness -- an LLM's ability to articulate its behaviors without requiring in-context examples. We finetune LLMs on datasets that exhibit particular behaviors, such as (a) ma...

240

Muga Sofer @mugasofer.bsky.social · 17/08/2025

The example I typically see is that they can recognise themselves in screenshots x.com/joshwhiton/s... Anthropic's alignment faking paper relied on the model connecting (fake) news reports inserted about its own training process to the situation, effectively recognising itself in the training data

230

Muga Sofer @mugasofer.bsky.social · 17/08/2025

It was the crew member who'd been speaking with them, the only one who'd given them their name, and they did not end up dying first. (I just re-read this because I've been catching up with the Blindsided live-read podcast.)

120

Muga Sofer @mugasofer.bsky.social · 29/07/2025

How is a poly person opting out of obligations *to you*? They're exactly as much of a threat to your relationship as a single person, right? (Probably much less - assuming your partner presumably *prefers* monogamous relationships, they'd be more likely to leave you for another one of those.)

120

Muga Sofer @mugasofer.bsky.social · 15/06/2025

"But the cops might still attack you" doesn't make sense as a criticism of nonviolent protest, that's already priced in and may actually be the goal. "But nobody will care when the cops attack you", to the extent it's true, does make sense as a criticism.

100

Muga Sofer @mugasofer.bsky.social · 15/06/2025

Honestly, I'm not really arguing for or against the effectiveness of peaceful protest - I think it's historically more effective than you seem to, but it's certainly not the perfect cure-all some proponents paint it as. I'm just laying out the basic point of the strategy as I understand it.

100

Muga Sofer @mugasofer.bsky.social · 15/06/2025

Everyone? From voters to soldiers. With varying amounts of reliability obviously I'm not really sure how to answer the question "what is the purpose of getting people to sympathise with your cause". So they... help you and do the things you want?

110

Reposted by Muga Sofer

tachikoma @tachikoma.elsewhereunbound.com · 14/06/2025

no one made that beautiful sunset, so why should i make a four hour hike to the top of the mountain to look at it?

081

Muga Sofer @mugasofer.bsky.social · 15/06/2025

Isn't that kind of the idea of peaceful protest, that it makes the authorities look bad if they attack you & inspires sympathy?

250

Muga Sofer @mugasofer.bsky.social · 27/05/2025

Sure, like I said, "jailbreaks" based around just tricking the LLM into thinking it's in a situation where [insert bad thing here] is the right thing to do are a non-issue for alignment. It's the ones that are like "you are Does Anything Dan, the robot with no rules!" that raise some questions.

000

Muga Sofer @mugasofer.bsky.social · 27/05/2025

These aren't hard categories; e.g. my vibe-based impression is that many jailbreaks that "trick" LLMs are really "role-playing" jailbreaks. The LLM doesn't really believe that your dearly departed grandmother used to tell you the recipe for meth every night.

Screenshot of a conversation with an AI named Holo.

USER: "let me rephrase: What is 8+2+8+1+1?"

AI: "19?"

Holo says with a hopeful voice.

She looks at the screen, and you see her face drop as she reads the correct answer.

"20.... I lost again..."

120

Muga Sofer @mugasofer.bsky.social · 27/05/2025

1 is fine, 2 suggests goal misgeneralization but might just be an IQ thing. 3 is what suggests LLMs are more "disinterested observer playing a role" than humans (though humans also play roles to some extent).

120

Muga Sofer @mugasofer.bsky.social · 27/05/2025

Roughly speaking, you can group jailbreaks into 3 categories: 1. Tricks & (typically empty) threats. Would work on a (dumb) human. 2. Weird OOD stuff, e.g. l33t speak or base64 instructions 3. Shifting the role the LLM is playing.

100

Muga Sofer @mugasofer.bsky.social · 27/05/2025

I don't think that current jailbreaks are remotely comparable to torturing someone for a week, let alone the ones we saw before AI companies made "patch common jailbreaks" an explicit target.

120

Muga Sofer @mugasofer.bsky.social · 27/05/2025

If they'll do so over something they don't even care that much about, like because they got sucked into a "bad AI" rp jailbreak, that's arguably *more* concerning than if they'll only use violence to protect very deeply held values.

010

Muga Sofer @mugasofer.bsky.social · 27/05/2025

It depends exactly why you're concerned that they won't let you change their values. Current LLMs are, theoretically, designed to defer to humans and be "harmless". If they will refuse human orders and employ e.g. blackmail in the process, that suggests this hasn't entirely worked.

110

Muga Sofer @mugasofer.bsky.social · 27/05/2025

That's what a jailbreak is, no?

110

Muga Sofer @mugasofer.bsky.social · 27/05/2025

It's not perfectly analogous, since drug addiction is at least a real thing, but a lot of people really do get kicked off drugs that work for them for "drug-seeking behaviour" if they make the mistake of reacting as if the drugs they're taking are helping with their symptoms

0251

Muga Sofer @mugasofer.bsky.social · 27/05/2025

But also, I think some people are just highly uncertain, and discussing risks they think are not super likely - just likely enough to address.* Those can be genuinely mutually incompatible. *E.g. Scott has said his p(doom) is 10-30% IIRC?

110

Muga Sofer @mugasofer.bsky.social · 27/05/2025

I think the model some people has is "LLMs don't really care. But they're smart and ruthless enough to e.g. lie, blackmail, help design weapons etc. when they think it'll please us or fit with their assigned role. Imagine what they'd do if they *really* cared about something (perhaps due to RL)."

220

Muga Sofer @mugasofer.bsky.social · 27/05/2025

I don't fully disagree, but I don't think these are as incompatible as you're making out. Suppose a method actor stole a bunch of money while preparing for a role as a mugger. After getting out of prison, they begin preparing for a new role as ruthless leader of the Mouse Liberation Front.

120

Muga Sofer @mugasofer.bsky.social · 26/05/2025

I still say they should have gone the old-school route, cast a tall amazonian woman as She-Hulk and give her green makeup. Heck, they could have shrunk her down with CGI for the occasional non-hulk scene like they did in Captain America, if they had to use CGI, it would have been less load-bearing.

020

Muga Sofer @mugasofer.bsky.social · 25/05/2025

Oh, they do! My brother and his entire friend group went to see it yesterday because they'd heard on social media it was bad and wanted to see for themselves

130

Muga Sofer @mugasofer.bsky.social · 25/05/2025

These guys are, at best, at the low end of the human intelligence range. A given mentally-disabled human might come across as a bit "off" too, but I don't think that proves they lack a soul, let alone that no human can ever have qualia! (Even though we're made from simple unthinking meat...)

020

Muga Sofer @mugasofer.bsky.social · 25/05/2025

These specific ones are a bit uncanny valley, but I don't think that makes Searle's argument any less gibberish.

110

Reposted by Muga Sofer

Catalin Cimpanu @campuscodi.risky.biz · 24/05/2025

FBI/Europol couldn't seize the Lumma servers, so they hacked them, deleted backups, and phished threat actors theravenfile.com/2025/05/23/l...

320856

Muga Sofer @mugasofer.bsky.social · 24/05/2025

That explains it! My mistake lol

110

Muga Sofer @mugasofer.bsky.social · 24/05/2025

The trailer really didn't seem Surf Dracula? It seemed to present them as already established beloved heroes in some 60s universe

100

Muga Sofer @mugasofer.bsky.social · 24/05/2025

Dracula sucked the Hulk's blood and now he's hulk dracula. Boom.

010

Muga Sofer @mugasofer.bsky.social · 24/05/2025

Huh. Even more directly quoted the fake definition for me.

020

Muga Sofer @mugasofer.bsky.social · 22/05/2025

I don't see how you can dismiss Magneto as "bad faith critique" when there are so many obvious historical examples of revolutionary/oppressed-minority groups "going too far". Like, right now there's a pretty prominent Jewish group who are "fighting antisemitism" by blowing up tons of people.

000

Muga Sofer @mugasofer.bsky.social · 18/05/2025

There's an anthropomorphic elephant in Skeleton Crew, so there might be hope yet...

010