Insider threats are overtaking external attacks as a top security concern, with AI accelerating the shift, new research from Exabeam shows.

According to the study, 64% of respondents now view insiders, whether malicious or compromised, as a greater risk than external actors. Generative AI (genAI) is a major driver, making attacks faster, stealthier, and more difficult to detect.

“Insiders aren’t just people anymore,” said Steve Wilson, Chief AI and Product Officer at Exabeam.

“They’re AI agents logging in with valid credentials, spoofing trusted voices, and making moves at machine speed. The question isn’t just who has access — it’s whether you can spot when that access is being abused.”

Insider activity is intensifying across industries, driven by both malicious intent and accidental compromise. Over the past year, more than half of organisations (53%) have seen a measurable increase in insider incidents, and the majority (54%) expect that growth to continue. Government organisations are bracing for the steepest rise (73%), followed by manufacturing (60%) and healthcare (53%), fuelled by expanding access to sensitive systems and data.

This surge is not uniform; risk trajectories vary sharply by geography and sector. Asia-Pacific and Japan lead in projected insider threat growth (69%), reflecting heightened awareness of identity-driven attacks.

The Middle East stands apart, with nearly one-third (30%) anticipating a decrease, a signal of either stronger confidence in current defences or a potential underestimation of evolving risks. These contrasts underscore the complexity of the insider threat landscape and the need for defence strategies that align with regional realities.

AI has become a force multiplier for insider threats, enabling actors to operate with unprecedented efficiency and subtlety.

Two of the top three current insider threat vectors are now AI-related, with AI-enhanced phishing and social engineering emerging as the most concerning tactics (27%). These attacks can adapt in real time, mimic legitimate communications, and exploit trust at a scale and speed human adversaries cannot match.

Recommended reading

• Security Leaders: Detection Alone Can’t Stop Cyber-threats
• Is Automation Fuelling a New Era of Cyber-crime?
• New Ransomware Gangs Drive Surge in Attacks

Unauthorised GenAI use compounds the challenge, creating a dual-risk scenario where the same tools meant to boost productivity can be repurposed for malicious activity.

More than three-quarters of organizations (76%) report some level of unapproved usage, with those in technology (40%), financial services (32%), and government (38%) experiencing the highest rates.

Regional variations are telling, in the Middle East, unauthorised GenAI is the top insider concern (31%), reflecting both rapid AI adoption and the governance gaps that can follow. Globally, the convergence of insider access and AI capabilities is producing threats that evade traditional controls and demand more advanced behavioural detection.

While 88% of organisations say they have insider threat programmes, most lack the behavioural analytics needed to catch abnormal activity early. Just 44% use user and entity behaviour analytics, the foundational capability for insider threat detection. Many continue to rely on identity and access management, security training, DLP, and EDR, tools that provide visibility but not the behavioural context necessary to spot subtle or emerging risks.