Google Cloud and Wiz debut AI security agents to counter automated zero-day exploits

Google Cloud unveiled three new AI-powered security agents at its Next '26 conference, positioning automated defense as the necessary response to AI-driven attacks that can exploit zero-day vulnerabilities within hours. The Threat Hunting agent proactively scans for novel attack patterns that bypass traditional rule-based defenses. A Detection Engineering agent identifies gaps in detection coverage and automatically creates new detections for threat scenarios. A Third-Party Context agent enriches security workflows with external data. All three are in preview and join an existing Triage and Investigation agent that Google says has processed more than 5 million alerts over the past year, reducing typical 30-minute manual analyses to about 60 seconds. Google Cloud stated that the core problem in AI security is that attackers are using AI models to find and exploit zero-day vulnerabilities within hours, while defenders are often still manually triaging alerts. The company is also one of the partners in Anthropic's Project Glasswing, which gives roughly 40 organizations access to Claude Mythos Preview for defensive vulnerability discovery. Wiz, which Google is acquiring for $32 billion, announced integrations extending its AI-Application Protection Platform to Databricks, AWS Agentcore, Gemini Enterprise Agent Platform, Microsoft Azure Copilot Studio, and Salesforce Agentforce. Wiz research found that 20 percent of vibe-coded apps contain security risks. Starting in May, Wiz will scan code directly inside the vibe-coding service Lovable before it ships. Wiz is also adding inline AI security hooks in IDEs, agent-based remediation that generates pull requests for vulnerabilities, and a dynamic AI Bill of Materials to inventory unauthorized AI coding tools across environments.