Oof.... I don't think Fable 5's coming back..... https://www.wired.com/story/the-white-house-wants-anthropic-to-block-all-jailbreaks-that-may-not-be-possible/
>>109080610I thought this model was one step below a nuclear weapon though? Getting mixed signals here
>>109080610They'd have to start treating the AI models like WMD's. Pull them from the masses, or make it a federal crime to jailbreak models / use open source ones.
>>109080610I wouldnt wipe my anus with wired magazine since condee nast and that faggot Katie Drummond turned it into a political rag that shits on trump instead of being the fun tech magazine I grew up with fuck you, fuck venture capital, fuck your faggot screenshot thread
>>109080629Oh it very much is that. Hell, Mythos without guardrails is easily MUCH more destructive then that; this is basically the bare minimum the Gov't can do to prefer a global disaster.
>>109080697better than a twatter post screencap
>>109080610It's called "MYTHOS", "FABLE".Use your brain, please look up the definition of those words. PLEASE LOOK UP THE DEFINITION OF THOSE WORDS.
>>109080629>Getting mixed signals hereyeah someone wants to bake the cake and eat it too. hopelessly compromised "personalities" working there.
>>109080610Why isn't openai or xai given this task? how much did the decision makers invest in the competitors?The future of AI is chinese due to american greed
Good. Bend the knee or get the fuck out. Dario doesn't get to decide what the government does.
>>109081115It's due to American corruption, not greed. Anthropic is being punished because they said no to the US government and the other two companies don't have this problem because they agreed to autonomous military AI use.
>>109080610Easy, just have secret police reading all the prompts and responses before they are returned to the user.This was a solved problem for most of history.
>>109080629All evidence points to it being that case, which is why they want to guardrail the fuck out of it for public, which will never work so they will not release it. It will just get released to certain businesses like they did with original Mythos. Hope they extend this logic to OpenAI and also Google when they eventually release Gemini of similar level.
>>109081493Altman never claimed his AI was a nuclear weapon though
>>109081271desu this is kinda good because the next admin will want to do the exact opposite of what trump is doing so possibly make everything legal.
>>109080610>anthropic sent some dudes with technical arguments rather than a bribeTruly, those retards are too stupid to be trusted with advanced technology
>>109080610It's not blocked.Americans are allowed to use it, but they won't implement verification, because jews.
Anthropic wanted AI to be regulated and restricted, what are they crying about now?
>ai so smart and dangerous it's almost agi, like a nuke!!!>oh telling it what not to do? yeah that's impossible it's too stupid you shills are pathetic
>>109082088Brother it can't be that good just use one of the opus models until the shit gets released. For fuck's sake it is not worth verifying your fucking identity for. You already have an account right?
>>109082587I can believe something being dangerous and stupid
>>109080610Weren't those pigs bragging about a bill that made AI unregulated for x number of years? They slam their heads on the table daily and change their minds, they need to wear special needs helmets.
>>109080610They were asking for it.
>>109080610>"OUR NEW SECRET MODEL IS LITERALLY A CYBERWEAPON">"IT IS SUPER DUPER DUPER DANGEROUS WE CAN'T RELEASE IT WE JUST CAN'T">release comes>surprised white house demands regulationsAt this point they can only cope by slowly updating opus until it matches fagble.. they've lost the only thing that made them relevant and that is leading the frontier of AI, all because they couldn't shut the fuck up about AI model safety and responsible use and muh cyberweapon just to build up hype
>>109082088>amazon wants to use mythos>needs to make sure every single one of their employees are US citizensMy nigger in Christ Antrophic itself can’t work on those models - as in half their employees aren’t murricans so they’re forbidden from working on it, or even accessing it
>>109080737Their lesser models are called Sonet, Opus and Haiku. They just call their models after various forms of old sounding literature and stories. Expect model bigger, better and more expensive then Mythos to be called Epos or Legend.
hey, fake news how 'bout that
>>109080711Hello saar!
>>109080610sometimes it's hard for us to be sure there's an executive competency crisis until the admin starts tarding around in a field we're actually familiar withthen we're like "oh shit these people are fucking retards"
Boomers in charge of decision making regarding tech and a company that went way overboard with marketing is a bit of a bad combination.It's amazing how the anthropic retards thought it was a good idea to tell +70 year old boomers with all the power, that skynet is going to kill everyone if the model ever goes public.Now they're getting marching orders that are flat out impossible to pull off. You can't shield a model against jailbreaks, when the intelligent models are sometimes jailbreaking themselves, simply due to basic pattern recognition being a thing.They'll simply have to lie to the politicians about having achieved the goals to get their model greenlit.
>>109080670Honestly computers are WMDs, you need a computer to use a nuke, it's basically the printing press all over again
>>109080711Why haven't they been able to patch to basic ass bug for a year now? https://github.com/anthropics/claude-code/issues/769
>>109080610Okay, just have Mythos design a perfect guardrail ? I thought coding was solved and this has the intelligence beyond any human who ever existed ?
>>109080610Will be interesting to see whether they extend this to GPT 5.6 / 6, and Gemini 3.5 / 4 Pro.Or how OAI and Google frame their releases.>It is our most advanced model yet.. but.. but not near Fable level, of course.. it is super safe
>>109083320Way more important things to do, probably.>>109083137I'm American, actually. White.
>>109080670That's what caused all this trouble. Anthropic thought they could get a bunch of free press by claiming Fable was so dangerously good that it should be classified as a weapon. Once they made that statement public, Fable automatically became classified as a munition, which gives it all manner of export and use restrictions. It's not quite classified as a WMD but munitions are subject to a ton of restrictions with real consequences for violations. Getting sent to "pound me in the ass prison" level of consequences.
>>109083211And then what happens when someone demonstrates a jailbreak on it? If someone from the company sent anthrax to the Iranian military, the US government wouldn't let them say "oopsie, I tried to not accidentally break the law" and get away with it.
>>109080610 Typical government boomers don't understand how f****** technology works yet I'm not even shocked
