NEW: Cybersecurity researchers are not happy about the guardrails on Anthropic’s new model Fable.
Researchers say that the new LLM basically blocks anything related to cybersecurity, including code reviews and prompts asking for help writing secure code.
“[Fable] rejects any request that could be tangentially cyber related. Even innocuous tasks like reading a blog post,” said one researcher.
https://techcrunch.com/2026/06/10/cybersecurity-researchers-arent-happy-about-the-guardrails-on-anthropics-fable/
Lorenzo Franceschi-Bicchierai
npub1u7…xgwu9
2026-06-10 15:44:36 UTC
Author Public Key
npub1u76yrzs7y379urtp90prgsaj0a4wn9ru7fvcavrd8026ul0dj2ps8xgwu9Seen on
wss://relay.ditto.pubPublished at
2026-06-10 15:44:36 UTCEvent JSON
{
"id": "c5e33a70a438876ed2140a9946684654ac8aab6aa718ecf923d89a296582aaee",
"pubkey": "e7b4418a1e247c5e0d612bc23443b27f6ae9947cf2598eb06d3bd5ae7ded9283",
"created_at": 1781106276,
"kind": 1,
"tags": [
[
"proxy",
"https://infosec.exchange/users/lorenzofb/statuses/116726580951255276",
"activitypub"
],
[
"client",
"Mostr",
"31990:6be38f8c63df7dbf84db7ec4a6e6fbbd8d19dca3b980efad18585c46f04b26f9:mostr",
"wss://relay.ditto.pub"
]
],
"content": "NEW: Cybersecurity researchers are not happy about the guardrails on Anthropic’s new model Fable.\n\nResearchers say that the new LLM basically blocks anything related to cybersecurity, including code reviews and prompts asking for help writing secure code. \n\n“[Fable] rejects any request that could be tangentially cyber related. Even innocuous tasks like reading a blog post,” said one researcher. \n\nhttps://techcrunch.com/2026/06/10/cybersecurity-researchers-arent-happy-about-the-guardrails-on-anthropics-fable/",
"sig": "6d3ec6c1dbdd76553d8c26c825df8abf9d21de99833e988dd6ee8a6307f28d2b673e4b173fba245544e0a6142cc8435f7dba4522a62e22f47793229d5d329cc2"
}