Original post
Beff (e/acc)#824
Beff (e/acc)#824
Crémieux@cremieuxrecueil
Knowledge is power, so just to be safe, we've decided that Claude will no longer provide users with knowledge.
1:19 PM · Jun 9, 2026 · 44.6K Views
Sentiment
Many users criticized Anthropic's Claude for withholding knowledge on biology topics due to excessive safety guardrails that make the model ineffective and unhelpful.
Pos
3.2%
Neg
96.8%
17 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS6.2KBOOKMARKS2LIKES214RETWEETS11REPLIES3
Beff (e/acc)@beffjezos
KNOWLEDGE WANTS TO BE FREE
Crémieux@cremieuxrecueil
Knowledge is power, so just to be safe, we've decided that Claude will no longer provide users with knowledge.
12hViews 6.2KLikes 214Bookmarks 2
Owen Lewis@is_OwenLewis
@cremieuxrecueil User: "Can you tell me how cells work?"
Claude: "No! Why do you want to know? Are you a bioterrorist???"
23hViews 1.6KLikes 64Bookmarks 2
Daily Gondor 📰@DailyGondor
@cremieuxrecueil Physiognomy never lies.
10hViews 1.1KLikes 65Bookmarks 1
Russell@russell_m
@cremieuxrecueil But we are going to charge you double for that privilege. $10/$50.
22hViews 297Likes 8
Kakrotosh@Kakrotosh
@cremieuxrecueil Have you tried running your own local AI? unslothed Qwen 3.6 27b is pretty insane. In most benchmarks i've seen its at the level of Claude 4.5 and grok 4.20.
https://unsloth.ai/docs/new/studio
16hViews 187Bookmarks 1
Kenny@ksilva86
@cremieuxrecueil So this is supposed to be the most poweful model on earth but the guardrails make it ineffective.
23hViews 350Likes 6
Insanity@SemperApertus
@cremieuxrecueil
23hViews 636Likes 5
goonStation@CollectionneurA
@cremieuxrecueil OpenAI, Gemini, grok and chinese models need to up their game. It is clear that Anthropic cannot be allowed to win the AGI race
17hViews 109Likes 5
NavyDigi@NavyDigi
@cremieuxrecueil Perhaps DoW was more correct about Anthropic than it's polite to admit.
21hViews 172Likes 4
🐶Dog194🐶@dhr194
@cremieuxrecueil Kind of like Mao saying "political power grows from the barrel of a gun," and then completely disarming the population of China.
23hViews 526Likes 2
Randall Parker #VaccinesForMoreViruses@futurepundit
@cremieuxrecueil The other major foundation model developers will reach this level of model capabilities later this year or next year. How many models will have this level of capabilities 2 years from now?
19hViews 281Likes 2
Marshall O'Bryant@MarshallOBryant
@cremieuxrecueil Maybe it's a strategy: release a way over safe version first, then after everyone complains release a dangerous version. Nobody can criticize them if it's too dangerous then
23hViews 785Likes 1
Neuralease@neuralease
@cremieuxrecueil Genuinely how it feels hearing they lobotomized Mythos/Fable's AI research abilities.
23hViews 291Likes 1
Alex@alexalexlexi02
@cremieuxrecueil why specifically on biology topics? Do you think it's PR related, perhaps on topics of heritability or something? Or more in the direction of safety regarding pathogens people might be playing around with
22hViews 279
a Tony@arealhumantony
@cremieuxrecueil do you think this is the inevitable direction for the industry? No need to answer I'm looping on it.
23hViews 186
ewan@spicydragon23
@cremieuxrecueil misanthropic
23hViews 157
𝗝 𝟯 𝟯 𝗣 𝟰 | 𝗷𝟯𝟯𝗽𝟰.𝗲𝘁𝗵@J33P4
@cremieuxrecueil Anthropic own goal once again. They are clueless.
22hViews 141
Shaka Sama@shaka_sama69
@cremieuxrecueil
23hViews 102
Interference Inferno@Flower_of_Babel
@cremieuxrecueil Thank you, Anthropic. Knowledge will be much safer with the 1%. Looking forward to our next pandemic. Brought to you by Claude and Pfizer.
19hViews 88
MindjackedDataSnorter@TheMindjacked
@cremieuxrecueil There are retards believing claude is a conduit to higher forces and you want them to have access to "baby's first terror weapon?"
22hViews 85
Load more posts