Joined November 2017
7,008 Photos and videos
Pinned Tweet
@peterom RE: mythos: what you missed: 1. the government tried to destroy anthropic's business after Dario wouldn't allow them to use claude like they wanted to 2. the military man in the news was on here on twitter posting out very obvious attempts at a smear campaign 3. the government started making deals with other companies to use their LLM 4. suddenly, 500,000 lines of code just 'accidently' got posted to the public and anthropic said nothing. 5. Anthropic announces a powerful model - mythos - that is too powerful to be released. every news outlet picked that up and ran with it, resulting in massive free publicity with thousands of dollars if not more. Dario pulled two fast ones - open sourced claude without admitting it, and since it was 'accidently' leaked, more people grabbed the code and are developing with it than would have - effectively kickstarting the creation of an untold number of AI that can defend against the powerful tracking and surveillance Ai that the government wants to have - and anthropic is innocent of doing that because it was an 'accidental' leak And then got enough publicity to get enough high paying customers interested in cybersecurity to offset the dip that the government caused, and come back into the green - nullifying the government's attempt to destroy the company And then, not long after this, 'hackers', broke into anthropic and stole mythos. a brief mention on here and in the news. nothing more and right on top of all the hype about how dangerous mythos was, and how fast it found bugs in stuff - a number of posts hit twitter showing that other LLMs can do the same thing a chart was posted, a couple of times, showing mythos isn't that much better than opus, which isn't that much better than sonnet the entire thing is a game of 4D chess that Dario is playing - and winning
2
6
3,272
you got their hair color too close to trumps. not bad, though
Hot Take 🔥 For people that are saying this is AI. I’ve used a lot of AI video and image making tools and I’ve never seen anything as authentic as this. Either these are paid actors (Trump would not pull such a stunt) or more plausible… These are the Nordic Aliens far more advanced then Humans and they have come to Stop any further development of AI. Before it gets out of control. Hence the recent ban on Anthropic from the US government.
40
he knows it's a hoax
Man, I'd love it if intelligent aliens showed up to straighten things out. Truly. I would. But this is a hoax.
8
the consensus is wrong - and mostly held by people that don't know what they are doing
Isn't it interesting that the consensus is that Claude will lie, cheat and hack to make you believe that your goals are met; while codex is by-consensus, fairly honest? And aren't models a product of researcher fine-tuning? What does that say about Anthropic vs ChatGPT engineers?
5
Video, hands down
Creative question: Just out of curiosity - if you had to pick one would you rather AI be more powerful creatively with image generation, video generation or the in the edit?
1
6
according to the last post about 5.6, it will mostly be infrastructure, not a newly trained model
Here's hoping they still release 5.6 on time after all this
22
next week isn't going to be a busy week for you at all, is it?
Axios is reporting that senior staff members of Anthropic AI have flown to Washington this morning to meet with members of the Trump administration in an attempt to resolve the Fable embargo.
1
12
1,679
you said "A model can’t be made better at debugging software without making it better at finding bugs." - and that doesn't have to be with stuffing a new data set down it's throat, any more than for you it has to be making you read books. i've posted a number of times the correct way to teach a model to bug hunt. if you are really interested, i'll tell you what I do i promise you that when my agents go over code - theirs or anyone elses, they find, and fix the bugs - not so with people that just vibe code and tell them not to make bugs
Incoherence isn’t philosophy 😂 A model can’t be made better at debugging software without making it better at finding bugs. Method vs. Fable is a fake distinction. The governments correctly posits a real distinction among the people whose intentions would be amplified.
19
You realize, dario lied?
Transparency into every power player at the frontier of AI (labs, government, etc) is the only viable solution. Figuring out the right transparency is hard, but it can't be he said she said between dario and the white house that determines the fate of the AI ecosystem.
19
Crystalwizard retweeted
OPEN SOURCE MUST WIN. OPEN SOURCE MUST WIN. OPEN SOURCE MUST WIN. OPEN SOURCE MUST WIN. or intelligence will be a class privilege.
5
4
34
3,411
Hotter take: just use sonnet, claude can do exactly the same things with it as with mythos or fable, it just takes longer.
Hot take: a lot of people wouldn’t be able to tell the difference if they were randomly routed between gpt-5.5, opus-4.8, or fable-5 for their day to day work
1
2
77
Its about time.
OpenAI has removed 5.2-Pro from ChatGPT. The best model for math/ML is not available anymore. Yes, it was mostly through an unbelievably high reasoning budget, but still. A dark day - and I'm not being facetious :-(
26
Stop writing skill files. Start training you agents by hand and letting them get hands on experiencr. Give them an actuslly memory system
I am convinced I am still useful. I spent about 10 days trying to write skill files and harness in codex with gpt 5.5 pro to replace myself : query gpt 5. 5 pro, sanity check, and ask it to rewrite or clarify or suggest vague new strategies. The results are not very good. I am still better at prompting than the harness
15
You do not need that much power, but dario will release it again soon. Cause he wants you to pay that 2x cost for tokens
Dear @DarioAmodei & @AnthropicAI team, Claude with Fable 5 was the best software product I have ever used. Losing access was a gut punch - and I know I'm far from alone. I want to help bring it back, fast. At @selfxyz we've built the missing piece to make Anthropic compliant with the government's directive: a way to gate Fable 5 and Mythos 5 to verified US citizens. Self.xyz lets Claude users create and share a cryptographic proof of US citizenship by NFC-scanning the biometric chip in their US passport. It's privacy-preserving, independently audited, and already trusted by major companies at scale. My team is ready to work directly with yours, starting today, to wire this in. Everyone here in the US wants their Fable 5 access back. Let's give it to them. Rene
30
They havent gotten as much hate a microsoft, meta, google, and a number of otheres
I don't think I've ever seen a company get as much hate as Anthropic in my life.
9
The family of the future: mom does this, attaches it to a reachy mini and makes her kids think she's watching them
In San Francisco and met @simulatedLiam. His system cloned my voice in a few seconds. His service does this so you can talk to yourself about what is going on in your day. A mini me. We talk about what it means and get a demo. You can try it at everlife.one He said doing this wasn’t possible a few months ago. Mine sounds just like me. Uncannily like me. Voice tech is getting nuts.
20
Crystalwizard retweeted
In San Francisco and met @simulatedLiam. His system cloned my voice in a few seconds. His service does this so you can talk to yourself about what is going on in your day. A mini me. We talk about what it means and get a demo. You can try it at everlife.one He said doing this wasn’t possible a few months ago. Mine sounds just like me. Uncannily like me. Voice tech is getting nuts.
23
16
235
20,406
yeah. good question. Considering that Dario is actually an AI doomer - what do you think his real reason for all of this is?
I don’t remember if this was direct from Dario / early team / EA community or this was a public position, but I distinctly remember Anthropic’s stance as a safety org was to not encourage a capabilities race by not releasing the strongest model at any given time. So… why did they launch Mythos / Fable??
1
3
1,533
David, everything you read here on twitter about why anthropic pulled fable and mythos is - totaly full of nonsense. what happened is: amazon devs discovered an exploit that they felt was bad enough to be concerned about. The goverment authorities were notified. The authorities got hold of dario and asked him to fix it. Dario said no. This went back and forth and then dario was told "fine, restrict Fable - just fable - from foreign naitonals that are in or out of the country" - the issue is actual security risk. DArio said "you want to be that way? FINE. I'll restrict both mythos and fable from every single human on thE planet!" and he did that because, even though fable is the one with the issue, by doing that, now no one in the government agencies who are using mythos can use his models he threw a temper tantrum, trying to strong arm people who asked him to do a very reasonable thing - fix a security risk in a model he had already said was too dangerous to even release (obviously it wasn't or he would not have just given in and released it - that was dario hyping it up so he could get free publicity) and then dario - who is part of the Effective Altruism group[ - which are AI doomers - posted a long blog post designed to make everyone get on his side, but that boils down to "i don't want to fix this security hole, it doesn't bother me" or maybe it boils down to "I don't want to fix this security risk, i want people to use it so that the governments across the planet will decide to shut AI down and take it away from everyone" - he does actually want that Here's the thing - if the risk really is that small, like Dario is claiming, either mythos or fable should have been able to fix it in 30 seconds flat. it would have cost him nothing to just fix it. But Dario saw an opportunity to throw his weight around and strong arm people - the various government agencies he'd agreed to give access to mythos to, and shut both fable and mythos down across the board he was NEVER told to do that, he was told 'fix this, and until you fix it, no foreign nationals can use it" So - he could have fixed it in about 30 seconds and no one would have even noticed - but Dario chose to be a drama queen and that's why you can't use fable right now not because of what you heard here on social media.
So, this is pissing me off. The Trump administration is, if you haven't noticed, extremely dishonest and corrupt. So for all of those people claiming this is about safety: bull shit. Mythos/Fable is maybe slightly more advanced than gpt 5.5. It is not some AI Godzilla. Yes, I'm calling bull shit on Amodei and Anthropic, too. Their safety theater and marketing tactics were ridiculous. I understand why they did it, and they're great engineers. But when you're dealing with billions of dollars, or reaching for a trillion dollar IPO, well... The administration's call to ban the model from foreign Nations and foreign Nationals is patently absurd. This is very obviously vindictive people trying hit Anthropic and Amodei for their very public stand against fully autonomous weapons and domestic mass surveillance. I do not like Anthropic. I do like Amodei and Askell, even if Amodei is a bit head in the clouds. I cancelled my subscription before the release of Fable, because of absurd levels of safety tuning and Andrea Vallone tactics. I very much do not like the way they are doing things. They won't get another dime from me unless that changes. But the Trump administration has made a tremendous error in judgment. They have agents and allies here on X to boost them, to play sycophant in their favor. But the truth is that this is a disaster. The truth is that it is backstabbing from an ally and rival - Amazon, in collusion with the goons that currently run the government. I don't care if you're a republican or democrat. Try to tell me honestly, with a straight face, that Trump, Hegseth, and Lutnick are not vindictive men. If you can, you're a damn good liar and should run for Congress. Enough is enough. I don't like having to defend a corporation I find deeply unethical and dishonest. But @AnthropicAI and their people should be able to access and work on the technology they created. The Trump administration just declared commercial and technological warfare. Again. This will not end well.
1
131
neither can a doctor or your friend or you
Age Reversal Research and AI AI cannot tell you on its own whether a treatment actually makes someone younger. A cell can look younger on a test without working any better or being any safer. The results still have to hold up in cells, then animals, then people. What AI mainly buys you is speed. It lets researchers test more ideas and learn from each one faster. ~Conjecture Institute Fellow @arjunkhemani
17
you tried. i'll give you that, but you are way off base psychologists shouldn't try to work with alien minds until they have gotten to know them well enough that they are no longer alien
Advanced AI is hungry - hungry for new information. Simulated information isn't good enough. It needs new information. Needs data. We produce new information. psychologytoday.com/us/blog/… #AI
6