Sophos AI Security 2026 report finds attackers are moving beyond experimentation and operationalizing AI for attacks
OXFORD, United Kingdom, July 22, 2026 (GLOBE NEWSWIRE) — Sophos today released its AI Security 2026 Report, finding that attackers are operationalizing artificial intelligence (AI) to collapse attack workflows from weeks to days. The report finds that AI’s most immediate impact on cybercrime is speed, as well as a rise in attacks using identity as the primary initial access vector (IAV), rather than inventing new attack types at this stage.
“Attackers still need initial access, still move laterally, and still exfiltrate through observable channels. What has changed is the clock,” said John Peterson, chief technology officer, Sophos. “For the first time we have observed AI being actively used as an operational force multiplier. While the tools and techniques were familiar, the speed of development, testing, and iteration was materially different. That is the AI threat that security teams need to prepare against. It means faster cycles and shorter windows to respond, with greater pressure on defenders to detect and contain activity before impact.”
Key Findings from the Sophos 2026 AI Security Report
- AI is compressing attack timelines and accelerating operational readiness.
- Enterprise AI identities, OAuth tokens, agents, APIs, and development tools are becoming high-value targets.
- AI-assisted social engineering and deepfakes are now operational tools.
- Threat actors are incorporating AI into underground markets, recruitment, prompt engineering, jailbreaking, malware development workflows, and criminal services.
- AI development infrastructure and supply chains are being targeted.
AI in the Hands of Attackers
In one of the report’s most significant findings, Sophos uncovered a campaign tracked as STAC6994, actively using AI to drive their operations; one of the first provable demonstrations of this happening.
The threat actor, was running a software development operation inside a customer’s network, and using approximately 12 AI agents to write and test attacks against endpoint agents, including Sophos, CrowdStrike, and Microsoft Defender. They produced nearly 80 modules and more than 70 evasion techniques, and turned what would have taken a human weeks into a few days. This dramatically accelerated the timeline of attack techniques reaching operational readiness.
This intelligence collection effort meant that we could stay ahead of the threat, defeating attacks before they made it into the wild
AI Identities Become a New Attack Surface
The report identifies enterprise AI adoption as the fastest-growing source of new exposure. As coding agents, assistants, and open-weight models take on privileged access to core systems, attackers are targeting the trust, credentials and access permissions surrounding these systems. As a result, AI identities, agents, OAuth connections, and API keys are increasingly becoming a high-value attack surface, and governance is not keeping pace.
Attackers are creating new pathways into enterprise networks by compromising OAuth tokens, AI service credentials, developer tools, and exposed AI infrastructure. This demonstrates that AI is now as much an identity, governance, and supply chain issue as it is a model security issue.
This is also reflected in the recent Sophos 2026 State of Ransomware report which showed that, for the first time in more than three years, identity has become the primary IAV.
AI-Powered Social Engineering and Deepfakes Become Operational
AI-assisted social engineering and deepfakes are making scams more scalable, more convincing across languages, and significantly cheaper to produce.
Incidents highlighted in the report include an AI-themed investment scam which drew a UK-based victim into a fake AI-powered investment platform through months of AI-themed lessons and coordinated messaging. The victim ultimately lost hundreds of thousands of pounds.
AI development infrastructure is also being targeted directly with attacks involving compromised developer tools and credential-stealing malware. Supply chain risks around model weights, training data provenance, MCP servers and inference infrastructure are also becoming more prolific.
“This report makes clear that AI security is no longer just about model behavior or speculative future risks. AI is actively being absorbed into criminal workflows and social engineering operations, as well as into enterprise software development and identity systems within legitimate organisations. That means the threat is in the here and now,” Peterson, Sophos. “As frontier models continue to advance, the next few months will be defined by how quickly organizations can govern AI use, secure the identities and connections around it, and keep pace with attackers who are capable of rapidly adopting new capabilities.”
The report is based on findings from Sophos X-Ops Managed Detection and Response (MDR) casework, SophosLabs analysis, Sophos Counter Threat Unit (CTU) intelligence, Sophos AI research, and endpoint and network observations across more than 625,000 customers worldwide.
Download the full Sophos AI Security 2026 report on Sophos.com.
About Sophos
Sophos, a global cybersecurity leader, defends more than 625,000 organizations worldwide with Sophos Fusion, the industry’s first and most complete AI-native cybersecurity defense system: a single, connected architecture where every control point operates as one. Powered by agentic AI and elite human expertise, Sophos detects, investigates, and neutralizes threats before they become business-disrupting events. Working alongside a global ecosystem of managed service providers, resellers, and technology partners, Sophos compounds intelligence from every threat encountered and every environment defended to make every customer’s defense stronger than the last. Sophos is headquartered in Oxford, U.K. More information is available at www.sophos.com.
CONTACT: Contact: Samantha Powers, sophos@walkersands.com
