Keep pulling the thread on Project Glasswing.
The US government's secret $9 billion budget for a CIA/NSA inference cluster, specifically to run Mythos-class models, demonstrates that advanced AI is now a critical component of national security strategy. This is further underscored by high-level negotiations between Anthropic and intelligence agencies, even amidst disputes with the Pentagon over usage guardrails.
Anthropic's Mythos model is in high demand globally, with European, UK, Canadian, and Asian governments seeking access. Simultaneously, Chinese firm Deepseek is raising $10 billion, backed by the Chinese government, and offering a powerful, low-cost alternative, suggesting a decoupling of AI ecosystems.
The rollout of Anthropic's Mythos model has dramatically accelerated vulnerability detection, with partners like Mozilla and Palo Alto Networks reporting 10x and 5x increases in identified issues, respectively. This has shifted the bottleneck from finding vulnerabilities to patching them, creating what Box CEO Aaron Levy calls a "security engineer boom."
The episode analyzes a fictional papal encyclical, "Magnifica Humanitas," which frames the AI debate around human dignity, value, and experience. The encyclical argues that AI, despite its computational power, lacks consciousness, embodiment, and moral understanding, and warns against reducing human worth to intelligence benchmarks.