"Claude can now be silently nerfed. Anthropic has decided it won't tell users when this happens." W T F!!
[flagged]
If Claude Fable stops helping you, you'll never know
361–370 of 534 posts
Re: If Claude Fable stops helping you, you'll never know
#362Earlier quoted context omitted.
Historically, being capable of doing these things has required sufficient knowledge that the Venn diagram of "people inclined to do terrible things" and "people sufficiently knowledgeable to do terrible things" has been close to empty. Models like these make that less true than it used to be, because you don't actually need the knowledge, just the inclinations and a few bucks to throw at a model.
Your "Venn diagram" is wrong. People don't decide against crime because they are dumb, they don't do it because of legal repercussions. Did you forget there's law? Why argue about dumbing down people in order to fight crime, that's nonsense. Private entities deciding to dumb down people as a replacement of law is worse than any crime.
I'm not primarily suggesting intelligence as a factor. I'm saying that among those who might want to do something especially harmful to humanity, it is exceedingly uncommon to, for instance, go study specific aspects of biology that would allow engineering a plague, in a long and diligent fashion without revealing anything, and still want to do it afterwards; that takes "premeditated" to a whole new level. And conversely, the kinds of people who study those aspects of biology in a long and diligent fashion aren't especially likely to have the temperament to decide they want to create a plague.
It's not that it could never happen. It's that it is much less likely.
Thought experiment: suppose there exists some trivial activity that would end the world, using everyday household objects that is easy to enact but vanishingly unlikely to do by accident, such that it could only happen if you made a deliberate choice to do it. For the sake of an absurd-but-clear information-theoretically-unlikely example, "write this exact ten-word sentence on a piece of paper, and place it in the microwave along with a vinegar-soaked match".
Now suppose that activity becomes public knowledge. How many minutes does the world last? I'd bet against more than a day (if betting were of any use).
Making it simple and widely accessible to do such things is a bad idea.
Re: If Claude Fable stops helping you, you'll never know
#363Just so everyone is aware. Anthropic has been sabotaging AI researchers and their codebases and shadow-nerfing accounts for several years at this point. This isn't new, but they hadn't disclosed it until now. Likely because it is getting to the point where it's too noticeable, or they're concerned about it leaking from employees.
Furthermore, the fact that they do these things, despite the incredible backlash... Just imagine what they're doing what your data and your IP.
Re: If Claude Fable stops helping you, you'll never know
#364Earlier quoted context omitted.
The moat is not the model, it's the harness. I wager that's one of the main reasons why Google made Antigravity closed source.
But harness is relatively easy to code yourself? They're just system prompt composer, with some tool functions that the LLM can invoke. I've vibe coded my own in just one day.
But is there anything preventing them from putting their own proprietary wolfram alpha/prolog/super duper expert system in there?
Re: If Claude Fable stops helping you, you'll never know
#365They have a silent nerfing system for their models and say so openly. The obvious question is how much it is being used already. Competitor companies being nerfed? Non Americans getting worse code? Punishing and rewarding users to maximize engagement, like online games do affecting victories through matchmaking?
All of the above, and it hasn't just started now. It's been happening for several years at this point.
Re: If Claude Fable stops helping you, you'll never know
#366Just so everyone is aware. Anthropic has been sabotaging AI researchers and their codebases and shadow-nerfing accounts for several years at this point. This isn't new, but they hadn't disclosed it until now. Likely because it is getting to the point where it's too noticeable, or they're concerned about it leaking from employees. Furthermore, the fact that they do these things, despite the incredible backlash... Just…
What evidence do you have of this?
Re: If Claude Fable stops helping you, you'll never know
#367Just so everyone is aware. Anthropic has been sabotaging AI researchers and their codebases and shadow-nerfing accounts for several years at this point. This isn't new, but they hadn't disclosed it until now. Likely because it is getting to the point where it's too noticeable, or they're concerned about it leaking from employees. Furthermore, the fact that they do these things, despite the incredible backlash... Just…
I guess it’s better they’re open about doing bad things. But now it’s a problem that they think this sets a precedent. They are one step away from feeling justified in using claude code running on a deepseek engineers laptop to hack deepseek to destroy their training system.
Re: If Claude Fable stops helping you, you'll never know
#368Earlier quoted context omitted.
It's such an obviously bad policy, it's mind-boggling that they thought this was a good idea. It just breeds paranoia and mistrust, especially when people are already a bit paranoid about silent model quantification for cost cutting reasons.
What's the alternative? Not release the model at all? "Make the guardrails better" isn't very hard and probably not worth the effort.
The alternative is to be explicit when you nerf, so users know what they are working with.
Re: If Claude Fable stops helping you, you'll never know
#369Earlier quoted context omitted.
What moat? There are multiple companies providing pareto-optimal frontier models, and it takes O(10) people to build one of these things. The rest is capital intensive, and the price will approach the cost of production over time. Thinking this is a profitable endeavor is equivalent to claiming coal plants have good margins because boilers are expensive.
Other models arent even close except for gpt 5.5. You're dead wrong on that. You read too many benchmarks and/or chinese propaganda. There hasn't been a serious contender in agentic SWE besides OAI and anthropic for a long time, and no chinese model has even reached opus 4.5 performance yet. The moat isnt insurmountable but it is very solid for at least a 12 month lead time. Which is such an insane amount of time in…
What's 12 months lead time worth? Not much from what I can tell. Contrary to what these AI companies might tell you, if an AI model can't do it, a human can still do the work.
Re: If Claude Fable stops helping you, you'll never know
#370OpenAI already did this when it released its "super scary advanced" security model. They silently return an earlier model's results if they think you're redteaming/abusing with it. https://openai.com/index/scaling-trusted-access-for-cyber-de...
They din't get as much pushback because they aren't the leader.