Comment on Claude 5 family's hallucinations look a lot like internal Anthropic emailsComments−cebert1moI tried this myself a few times from Claude.ai, but was unable to produce a similar result after many attempts.−arm32OP1moAnthropic has patched the `---` trigger roughly 30 minutes ago, by all reports.−saidnooneever1moi hope thei asked the LLM to quickly patch it so now there is something else as stupidly broken -_- my god these guys are rubbish at engineering. its like they really need some tool to do it for them o.O so weird
Comments
I tried this myself a few times from Claude.ai, but was unable to produce a similar result after many attempts.
Anthropic has patched the `---` trigger roughly 30 minutes ago, by all reports.
i hope thei asked the LLM to quickly patch it so now there is something else as stupidly broken -_- my god these guys are rubbish at engineering. its like they really need some tool to do it for them o.O so weird