Episode Details

A tumultuous period for Anthropic, specifically concerning its AI coding assistant, Claude Code. A major focus is a massive source code leak on March 31, 2026, which exposed over 512,000 lines of proprietary code and led to a "clean-room" rewrite by the developer community. Simultaneously, users have reported critical bugs that cause token consumption to inflate by 20x and a noticeable regression in reasoning capabilities for complex engineering tasks. Anthropic has responded by limiting usage quotas and testing new pricing tiers, sparking a backlash from developers who feel the tool's reliability and accessibility are declining. Despite these challenges, the company continues to release rapid software updates and preview next-generation models like Claude Mythos to regain trust. These events highlight the legal and technical vulnerabilities inherent in the rapidly evolving AI landscape.