Elon Musk, Mark Zuckerberg, and David Sacks killed Trump's draft AI safety executive order through three Wednesday-night phone calls. The tech leaders opposed the regulatory framework before it could advance, marking a direct intervention in White House policy by major AI figures.

The move came as Anthropic closed a $30 billion-plus funding round, cementing the startup's position as a heavyweight in frontier AI development. Meanwhile, Microsoft quietly terminated its internal Claude Code pilot after token billing consumed the entire annual AI budget for the program. Microsoft redirected affected developers to Copilot instead.

Security issues mounted across the sector. CISA documented 15,000 attacks exploiting a Drupal SQL vulnerability within a single week. More alarming, researchers discovered TrapDoor, the first cross-registry supply chain attack striking npm, PyPI, and Crates.io simultaneously. The attack used .cursorrules and CLAUDE.md configuration files as infection vectors, targeting developer toolchains across multiple ecosystems.

The White House personally intervened to override Pentagon decisions, ensuring Claude remained operational inside the NSA. This move contradicted the security establishment's original position on the AI model's deployment in national defense infrastructure.

The events reveal sharp tensions between tech leadership and emerging regulatory frameworks. Musk and Zuckerberg's direct intervention suggests they view Trump-era AI policy as threatening to their competitive interests. Anthropic's massive funding round demonstrates investor confidence despite heightened regulatory scrutiny. Microsoft's Claude pilot cancellation highlights practical budget constraints in enterprise AI spending, particularly as token-based pricing models scale unpredictably.

The security incidents underscore vulnerabilities in open-source AI tooling supply chains. Configuration files designed for AI workflows became attack surfaces, suggesting developers and platforms need revised security practices for AI-native tools. The NSA precedent for Claude deployment signals confidence