As the geopolitical display around AI unfolds, it’s going to be more critical than ever to have transparent, auditable agents. Claude Code has been caught fingerprinting when you connect to Chinese servers. Interestingly, people distrust Chinese models, which can be self-hosted in controlled, even air-gapped environments, but don’t distrust the closed labs. Here Anthropic is actively obfuscating this fingerprinting. So much for their ethical approach to AI. vincentschmalbach.com/claude-code-china-router-fingerprint

My team and company specifically sought out and discovered openhands after Fable 5 got pulled. That spooked us and we will always remember how fast and easy these tools can get pulled.