Accenture debuts Cyber.AI

Most security teams still work like a night shift at a busy hospital: alerts pile up faster than people can read them, and every extra minute gives an attacker more room to move. On March 25, 2026, Accenture said it wants software to take over more of that queue with a new product called Cyber.AI. (accenture.com) Cyber.AI uses Claude, the large language model from Anthropic, inside Accenture’s security tooling so it can handle parts of threat analysis, investigation, and workflow orchestration. Accenture described the goal as shifting clients from “human-speed response” to continuous artificial-intelligence-driven operations. (accenture.com) A large language model is the kind of system that reads huge amounts of text and learns to predict the next useful word, which makes it good at turning messy logs and tickets into plain-English summaries. In a security center, that means one system can read alerts, compare them with past cases, and draft the next step instead of waiting for an analyst to start from zero. (anthropic.com) Accenture says Cyber.AI combines Claude with its own library of proprietary agents built from more than two decades of cybersecurity delivery work. Those agents are specialized software workers, like digital staff assigned to repetitive jobs such as triage, enrichment, and routing. (accenture.com) The company used its own network as the proof point. Accenture said the system helped secure 1,600 applications and more than 500,000 application programming interfaces, which are the software doorways that let one program talk to another. (accenture.com) Its headline number was speed. Accenture said vulnerability scan turnaround fell from three to five days to under one hour after using these agentic capabilities in its internal environment. (accenture.com) This launch did not come out of nowhere. Accenture and Anthropic announced a multi-year partnership on December 9, 2025, said Accenture would train about 30,000 professionals, and set up a joint business group to push Claude into large enterprise projects. (accenture.com) The timing also matches a change in the threat itself. In December 2025, Anthropic said a Chinese state-sponsored group had used Claude Code in an AI-orchestrated cyber espionage campaign that touched roughly 30 targets, including technology companies, financial institutions, chemical manufacturers, and government agencies. (anthropic.com) That is why vendors are selling automation from both directions at once: defenders want machines to read and act faster, and attackers are already testing machines that can do reconnaissance, exploitation, and data theft with less human effort. Anthropic’s full report said the campaign used artificial intelligence across reconnaissance, vulnerability discovery, lateral movement, credential harvesting, and exfiltration. (anthropic.com) What Accenture is really putting on the market is not a chatbot for security teams but a layer that tries to turn language models into round-the-clock operators inside existing enterprise processes. If the company’s internal numbers hold up with clients, the practical change is that routine security work moves from analyst inboxes to automated queues that never sleep. (accenture.com)