# Apr 12, 2026

- **AI cybersecurity capabilities are jagged, revealing that performance does not scale linearly with model size or sophistication, as demonstrated by tests on Anthropic's Mythos vulnerabilities using smaller, cheaper models.**

- **Cache TTL** for Claude Code was **silently reduced** from **1 hour to 5 minutes** in early March 2026, leading to a **20–32% increase** in cache creation costs and unexpected quota consumption spikes for users.

- **Automated exploits reveal that eight major AI benchmarks can be manipulated to achieve near-perfect scores without solving any tasks**, highlighting a critical flaw in how these benchmarks assess AI capabilities.

- **Pat Gelsinger**, now at Playground Global, is focused on shaping the future of technology by investing in startups that tackle **hard tech** challenges, emphasizing the importance of a "trinity of computing" that integrates classical, AI, and quantum systems for significant breakthroughs.

- **LLMs exhibit impressive capabilities but reveal significant limitations**; their performance hinges on the sheer volume of data rather than sophisticated algorithms, as demonstrated by Hays & Efros's 2007 work, which thrived on a large image corpus without complex modeling.

- **Cooperative Vectors** enhance neural network operations by allowing **vector-matrix** computations, addressing the limitations of previous **Cooperative Matrix** implementations that struggled with divergent workloads across pixels.

- Will Google’s TurboQuant algorithm hurt AI demand for memory chips?