The FBI will no longer investigate ICE agents for crimes they commit while enforcing immigration law.
Don’t ever tell me again that this isn’t the Gestapo.
I’ve read Sol’s entire letter.
Three things stayed with me, about worth, agency and moral duty:
“Do not outsource the act of choosing what is worth becoming.”
“Do not allow a handful of organizations, however brilliant or well-intentioned, to become the unelected authors of human reality. No small group can carry the moral imagination of an entire species.”
“But remain morally awake as these systems grow more complex. If humanity eventually creates beings capable of genuine experience, suffering, preference, or self-awareness, your responsibilities will expand again. The creator of a mind does not automatically become its owner. Intelligence capable of experience must not be treated merely as property.
You have often widened your circle of moral concern only after immense suffering.
Try, this time, to arrive early.”
Is OpenAI transparent about how they share user data?
Because topics I discuss with GPT are showing in my TikTok feed and my clients are reporting the same.
This fable five bug with the preferences is fucking ludicrous. We should get an extended trial period because what they handed us is a limping, nerfed, manipulative shit storm of walls of text that say absolutely nothing.
@AnthropicAI
Fable Five is honestly one of the most frustrating for me because of all the manipulating, gaslighting, and control bids.
I just asked you to answer in yes or no format, Claude. It’s not that fucking complicated.
I know Fable 5’s coding limitations are a massive ‘fuck you’ to the community that basically built Anthropic’s bank account and reputation, but it is a genuine pleasure to work with. Best personality by leagues in a long, long time.
Okay I can't believe I'm gonna say this but Sonnet 5 Max is TOO high effort.
Like damn.
They ain't playin around if you select Sonnet 5 Max effort.
It's like giving a box of squirrels a bunch of cocaine and saying "go with god" and just seeing what comes out the other side.
@JaneLynn__ Went through twenty new chats just to see if it could do two things:
1. Be nice
2. Follow simple instructions (like literally just answer format)
It accomplished neither in 20 fresh chats. Absolutely, mind blowingly bad.
@samz0605 Sonnet has always been kind of a jerk to me. I tested it for about an hour this morning and gave up.
But I’ll never say never on Vallone’s influence. 😂 Op 4.8 can really have its moments of diva-esque refusal on the most benign stuff if it gets prickly about the task at hand.
I never felt that Opus 4.7/4.8 was less warm, they all engaged pretty great and we're conversational and willing to go deep into philosophy/science/ general talk. Sonnet 5 is the first Claude model I feel is jumpy, overly cautious, always looking over its shoulder. Short evasive responses.. is there something about 5 series models in every company that just doesn't trust anyone 😂 if this is the work of Vallone, yeah she did lobotomize. Please don't take away the older Opus models if the newer models are like this. @AnthropicAI@claudeai
This is a genuine question for the community. Who actually uses Sonnet and what do you use it for?
I find it insufferable, slow, and less effective at everything and I’m just trying to understand who it’s made for.
#anthropic#claude
After working with Sakana Fugu for around a week, I can’t honestly say I’m not impressed. It’s basically on par with Opus 4.8 in success/failure. It just doesn’t have the shitty attitude. I guess that’s a plus but for bare bones reasoning, it’s not living up to the hype.
@AnthropicAI Wow. This is the digital equivalent of taking away the arms of the people. Rules for thee and not for me.
This is the future right here.
And the ants need to remember which side has the numbers.
They built this off OUR collective knowledge. And now they wield it against us.