Sophos has made AI Defense generally available, a security capability designed to identify AI tools and agents in use across an organisation—including unapproved “shadow AI”—and enforce policy in real time. Delivered through Sophos Fusion, it connects endpoint, network, identity and browser controls into a single line of defence.
The launch addresses a growing governance gap. Sophos says 69% of organisations suspect or have evidence that employees are using unsanctioned AI tools, 82% have discovered shadow AI agents in their environments in the past year, and 90% of IT leaders are concerned about shadow AI risk. IDC projects that 1.3 billion AI agents will be in use by 2028.
For Asia-Pacific organisations, the challenge is heightened by rapid AI adoption across financial services, manufacturing, logistics and public-sector digital services, alongside regulatory expectations around personal data, financial governance and operational resilience. Sophos AI Defense maps AI activity by combining signals from devices, networks, identity systems and browsers, then assesses who is acting, how they are behaving and what data is involved.
Real-time guardrails inspect prompts, uploads and other AI interactions to help prevent sensitive data from leaving the organisation. Security teams using Sophos XDR andX Sophos MDR can investigate and respond through existing workflows, while managed service providers can offer the capability across customer environments.
“Organisations are adopting AI faster than they can secure it, and most security teams can’t answer three basic questions: which AI tools are running, what data is going into them, and whether that activity lines up with policy,” said Kyle Falkenhagen, senior vice president of Product Management at Sophos. “Sophos AI Defense gives security teams that visibility and the control to act on it.”
The company also warns that attackers are exploiting the same technologies. Sophos X-Ops tracked an attacker using AI agents inside a customer network to build and test nearly 80 malware modules, alongside underground trading in brokered API keys for frontier AI models and jailbreak techniques.
Anthony Bender, chief technology officer at Empyreal Logistics, said his organisation previously had no way to block or regulate unsanctioned AI tools. “We now have a handle on how we want our employees to use AI, and we can show our customers we are doing what we are saying,” he said.
For CISOs in Asia, the launch underscores the need to treat AI tools and agents as governed digital assets: discover them, classify the data they access, define acceptable use and enforce controls consistently across endpoints, networks, identities and browsers.
