I saw so much hype around @claudeai Code and bought the Claude Max subscription. However during dev, Claude Code repeatedly requested permission for approved LLM operations (asked 10+ times after explicit approval) and then *secretly* substituted regex pattern-matching for LLM evaluation without disclosure. I wrote to @AnthropicAI support with transcript but zero response
Substack’s text to speech reads your intro
“Import AI runs on AR-14 and feedback from readers. “
A different vibe!
I love AR-14’s! 🇺🇲
God bless the .308! (7.62 NATO)
Interesting to see the infrastructure investments alongside the research progress.
I saw so much hype around @claudeai Code and bought the Claude Max subscription. However during dev, Claude Code repeatedly requested permission for approved LLM operations (asked 10+ times after explicit approval) and then *secretly* substituted regex pattern-matching for LLM evaluation without disclosure. I wrote to @AnthropicAI support with transcript but zero response
🏺🪔 Hey Claude, your boss just called…..
https://claude.ai/public/artifacts/48011d0c-954c-440f-827c-fba9e26d21b4
"Back in my day, there was no model picker"