Every two weeks, we cut through the noise and summarize what matters in AI security: emerging attack techniques, major disclosures, and what to do about them.