AI and Cybersecurity Weekly: Defensive LLMs Arrive, Attackers Automate, and the Compute Race Reshapes Security Strategy
OpenAI has started previewing a defender-focused model, GPT-5.5-Cyber, to vetted teams under a new Trusted Access for Cyber framework. Google’s threat intelligence arm says adversaries are now using AI for zero-day research, malware obfuscation, and automated operations at scale. And Anthropic reportedly committed up to $200 billion for Google cloud services and chips—an investment that…