Rest of World: Hirundo and Lazarus Scrub Chinese Model Censorship, Deploy 'ReAligned' Qwen at ~12 US Firms

Israeli-founded Hirundo and Boston's Lazarus AI are using machine unlearning and weight edits to strip CCP-aligned censorship and harden security on open-weight Chinese models like Qwen and DeepSeek. Lazarus has already deployed its 'ReAligned' Qwen variants at roughly a dozen US enterprises using leaked guidelines and Gemini Flash outputs as the realignment signal. CrowdStrike research cited in the piece found DeepSeek-R1 ships code with severe vulnerabilities 50% more often when prompts touch sensitive political topics.