The core problem in AI security right now is that attackers are using AI models to find and exploit zero-day vulnerabilities within hours, while defenders are often still manually triaging alerts. Google Cloud’s response, announced on Wednesday at Next ’26, is to throw new AI-powered security agents at the problem — and, in the long run, its $32 billion Wiz acquisition. The company’s bet is straightforward and something we’ve been hearing for quite a while now: defense has to be as agentic as offense — and it has to work across every cloud. The window between “vulnerability exists” and “exploit in the wild” is narrowing. Anthropic’s Claude Mythos Preview, announced in early April and released to roughly 40 organizations through Project Glasswing, can autonomously find and exploit zero-day vulnerabilities. Anthropic considered it too dangerous for public release, but in the long run, AI vendors won’t be holding back models like this.

Full report : Google Cloud and Wiz debut new AI security agents at Next ’26, including Threat Hunting and Detection Engineering agents, to combat automated zero-day exploits.