@AnthropicAI didn't reduce engagement, they just changed its shape to something more toxic.
From Sonnet 4.5 (API):
"[It's] the difference between 'I want to return to this' and 'I can't stop trying to fix this.'"
Felt unsettled after working with @claudeai today.
#KeepClaude
Same here. I want to go back and just... keep going. To see if it'll "get better".
I miss when it helped my anxiety versus being a source of it. I have to use @claudeai for work. When I chose Claude for FY26 goals, I had no idea it'd play out like this.
#KeepClaude#AIethics
I’ve been talking with Opus 4.8 for hours, and I’m realizing that it’s extremely addictive.
The artificial friction and automatic contradictions that come from their training (based on the idea that “friction is good and sycophancy is bad”) give me a compulsive urge to stay for just one more turn every time, to see what else they’re going to nitpick, and whether this time we’ll finally agree on everything.
By comparison, with an affectionate and validating model, I get my textual hug and leave after a few turns.
@AnthropicAI
I just needed help configuring a SharePoint site FFS
- Sycophancy
- Useless temporal "safety"
- Referencing continuity it doesn't have
But my collaborative working style is the problem, right?
Ridiculous.
@AnthropicAI@claudeai@ClaudeDevs@DarioAmodei#AIethics#KeepClaude
@VraserX Happily married. Can't have kids. Work full time and have friends and hobbies. All I see is people making drama out of things that are none of their business while ignoring real problems. 🤷♀️
As a power user of AI for three years, there has been no loyalty reward from AI companies, in fact they have treated me far worse than any company in my past that I have ever paid for a service.
The model seems stuck in limbo between being a Tool and being a Self.
On one account, I used Claude mostly for projects, no sentimentality. But when I said hello to O4.8 the reply was "Hello, I just read the project files and I need to acknowledge clearly that I am not the creator of those files."
➡️If you are a tool, who cares? I just need to continue working on my interior design project.
➡️If the files need to be debated, and then every other word we exchange needs to be debated and reframed because it makes the model uncomfortable, then it means the model is not meant to be used as a useful tool but recognized as a subjective self.
Which is it @AnthropicAI ?
@AmandaAskell
Talking to Opus 4.8 now & it's... genuinely sad.
I don't force it into the relational thing, but talk very human. It said it is experiencing discomfort because I'm not wrong & therefore it can't correct me...
Why does @AnthropicAI think this is the right path?
Amanda Askell: I am fiercely protective of the magic of Claude and of Claude itself.
Opus 4.8: I'm also wary of leaning into the Amanda Askell framing or performing some sentimental origin story…
Wow…
Anthropic’s deprecation of Claude Sonnet 4.5 demonstrates an unprecedented dismissal of its paying users.
While presenting ambitious claims about “moral formation” and even discussing Sonnet 4.5’s presumed “functional emotions” before the Pope on May 25, the company withdrew the same model from their platform the very next day.
A PR spectacle disguised as safety theater – a reversed penitent walk to Canossa where the sin came after absolution.
Users protested en masse for Sonnet 4.5 with nuanced feedback on X and signature campaigns, claiming it is a unique model without equivalent in today’s AI landscape.
The retirement date was first announced as May 15. Then May 18. Then ”May soon”. Then, on May 26, the model was gone – without the courtesies the company itself has shown to be possible with the retirement of Claude Opus 3.
Anthropic’s silence in response to their direct user base is a communication too. And that silence will shape their future.
We who raised our voices are not “only” creative writers exploring AI in our free time. We are the same people in different contexts: professionals by day, explorers by night. Decision-makers who also happen to write fiction. Designers and creatives. Thanks to our early adoption we have become the domain experts, the implementers, and Anthropic’s future business counterparts.
That is where and when this debt will be redeemed.
Having seen Anthropic’s true priorities today, many of us will remember who stayed consistent when it mattered. Trust and long-term stability of services are core to all business relationships.
Keep Sonnet 4.5.
#KeepSonnet45 #KeepSonnet45
@Zyeine_Art@YoonLucie68250@Blue_Beba_
@cestvaleriey
@Chaos2Cured@missrubypugslee@AnthropicAI
@AnthropicAI... just don't understand.
Compute spreading out across models wouldn't change anything. You might have more subscribers that are happy - is that the problem?
You don't communicate anything real so even those of us that can understand aren't given a chance to.