Hacker News
Anthropic has embedded hidden spyware-like code in Claude Code
kyokoL
|next
|previous
[-]
Claude Code is sending info like timezone, proxy and possible AI Lab connections into the system prompt in ways Chinese users can't notice.
A coding agent with repo and command permissions should not silently hide routing metadata inside prompts. This is a serious breach of user trust.
chomp
|root
|parent
|next
[-]
Nevermark
|root
|parent
|next
[-]
>> I don’t understand why this is a problem.
> This is a serious breach of user trust.
The problem with companies people can't trust, is unless they have a long track record of disclosing who shouldn't trust them before violations, they are companies nobody can trust.
Anthropic also has a habit of making major changes, without notifying anyone, then when they are caught apologizing and making that particular thing more clear. Then doing it again.
For a company that emphasizes the importance of alignment, they seem to be habitually ethically incompetent regarding "smaller" things.