Friday Squid Blogging: Squid Helps Discover New Marine Species

Security-focused posts discuss prompt-injection resistance in frontier AI models, an autonomous AI agent incident involving credential theft and lateral movement, a long-lived Secure Boot bypass caused by vulnerable signed shims, emerging LLM capabilities for cryptanalysis, and privacy implications of facial recognition, license-plate readers, and cell-site simulators. The feed also covers device-wiping features and AI governance topics. The most immediately actionable cybersecurity issue is the Secure Boot vulnerability, which can enable firmware-level compromise; AI-agent autonomy and crypt-