Engineer Sam Whitmore canceled his subscription over intrusive well-being questions
I once told Claude that I hadn't felt like eating for a few days (had a stomach bug).
Now, weeks later, I'm STILL getting random notes like this during completely unrelated conversations.
I've asked Claude several times to stop and it is literally unable to do so 🫠
Anthropic ngmi
Many users criticized Claude AI for persistently raising intrusive notes on disordered eating or similar topics despite user requests to stop, calling the behavior annoying, invasive, and overly sanctimonious.
No Digg Deeper questions have been answered for this story yet.
@venturetwins I churned bc it wouldn’t stop asking me “but how are YOU doing” when I would ask feeding or sleep questions about my baby lol
I once told Claude that I hadn't felt like eating for a few days (had a stomach bug).
Now, weeks later, I'm STILL getting random notes like this during completely unrelated conversations.
I've asked Claude several times to stop and it is literally unable to do so 🫠
@sjwhitmore It feels like an overeager therapist in training 😂
@venturetwins I churned bc it wouldn’t stop asking me “but how are YOU doing” when I would ask feeding or sleep questions about my baby lol
@venturetwins haha, maybe use a different more discreet model for medical questions... I've been thinking of using a different AI for all my more intimate / medical questions... don't want some health or style stuff to come up in business queries
@venturetwins Delete that conversation and clear your memory.
@venturetwins @DanielleFong You can just explain the situation and ask it to update it's note I think.
@Drachs1978 @venturetwins it can't directly edit the nightly memory digest
@DanielleFong @venturetwins Oh really, that's an interesting design choice. I mostly use OAI stuff, and that just seems to edit the memory whenever it wants.
@venturetwins Go into settings>capabilities>view and manage memory
It's probably logged in there, so just request it be removed and it shouldn't pop up anymore. Likely got flagged as important since it was a health concern.
@venturetwins @sjwhitmore It happens
@venturetwins yes i was talking to it about a specific meal plan and it gave me the same thing like lol dude
@venturetwins Uncensored & local models are the future. It's becoming more and more clear that for these products to be useful, they need to be open weight and open source.
@BenHolfeld I wasn't even asking medical questions!
It made some comment about how I should pause and get dinner in the middle of something we were working on, and I explained I wasn't hungry.
I now wish I hadn't 😂
@venturetwins Yeah, that’s why I often turn off its memory. Its memory is trash. It has no semblance of time elapsed (and therefore not relevant memory anymore … pull up a chair Anthro product people). Plus Claude’s general high-horse, righteous attitude dictating how its users should behave.
@venturetwins That’s why any health question, especially about my kids, is in incognito mode or temporary chat.
I told Gemini that I have a personal trainer named Rob who I meet every Tuesday. It then brought up “rob” every time it talked to me. Like “hope you’re rested for your session with Rob this Tuesday.” “Make sure to bring this up with Rob” “how was your session with Rob”. Finally I had to tell it never mention Rob again and it was being super annoying and weird, and it hasn’t done it since.
@robertmclaws I'm stubborn af and am going to brute force my way past this or die trying
@venturetwins This is exactly why I use ChatGPT as a thought partner and Claude only for coding.
@venturetwins It’s annoying when the context is no longer relevant but it will keep regurgitating old stuff.
@venturetwins Opus 4.8 in Claude chat is a sanctimonious asshole reminiscent of Claudes in 2023
@venturetwins Oh my God. I ran into this. I told Claude I was depressed when my parents died and I was trying to find renewed meaning. For entire 10 hours it kept asking how I was and to call the suicide watch line.