Survey from TrendAI™ finds pressure to deploy AI for efficiency outpaces governance, transparency and accountability
STOCKHOLM, 25 March 2026 – Global cybersecurity leader TrendAI™ today published a new study showing that companies worldwide continue to rush AI adoption despite well-known security and regulatory risks.
The survey, which polled 3,700 business and IT decision-makers, found that 67 percent feel pressured to adopt AI even when they have security concerns. One in seven respondents described their concerns as “extreme” yet proceeded because of competitive pressure.
“Companies are aware of the risks but lack the means to manage them,” says Martin Fribrock, Country Manager Sweden, Finland and Baltics at TrendAI. “When AI deployment is driven by competitive pressure rather than governance maturity, organizations integrate AI into critical systems without the controls needed for safe operation. This study highlights the importance of helping companies deliver strong business outcomes with AI while maintaining robust risk management.”
The risks tied to hasty AI adoption are compounded by weak governance and unclear accountability for AI-related risks. Security teams often find themselves reacting to top-down decisions, which can lead to ad hoc measures and increased use of shadow AI and other unauthorized tools within organizations.
Recent research from TrendAI™ also documents a clear shift in attacker behavior: adversaries are already using AI to automate reconnaissance, speed up phishing campaigns and generally lower the barrier to cybercrime. That trend increases both the speed and scale of attacks.
Governance lagging behind AI use
Organizations are adopting AI faster than they can secure it, widening the gap between ambition and control. 57 percent report AI deployment progressing faster than they can secure usage, and more than half (64 percent) say they only have a moderate understanding of the legal frameworks that govern AI.
Governance maturity remains low. Only just over a third (38 percent) say they have comprehensive AI policies in place, while many are still developing them. 41 percent cite unclear regulations or compliance requirements as a barrier. In practice, this means AI is being introduced without fully established rules for safe and compliant use.
Trust in autonomous AI remains limited
Confidence in advanced, autonomous systems is still maturing. Fewer than half (48 percent) believe AI agents will significantly improve cyber defense in the near term, and concerns about data access, misuse and loss of control persist.
The data shows where concerns are greatest. More than four in ten companies (44 percent) identify AI agents accessing sensitive data as the top risk. Over a third (36 percent) warn that malicious prompts could compromise security, while one in three (33 percent) note an expanded attack surface for cybercriminals. The same proportion (33 percent) fear misuse of trusted AI tools and risks tied to autonomous code distribution.
Nearly a third (31 percent) say they lack visibility or traceability in their systems, raising serious questions about an organization’s ability to control or intervene once agents are operational.
About 40 percent of companies support implementing AI-based emergency shutdowns (kill-switches) to stop systems in the event of failure or misuse, while almost half remain unsure. This lack of consensus highlights a deeper problem: organizations are moving toward autonomous AI without agreement on how to maintain control when it matters most.
“Agentic AI introduces a new class of risk for businesses,” says Martin Fribrock. “Our research already surfaces clear concerns, from exposure of sensitive data to losing operational control. Without transparency and governance, companies implement systems they neither understand nor can steer. That risk will only grow unless we act.”
About TrendAI™
TrendAI™, a business unit within Trend Micro and a global leader in AI security, helps make digital information exchange safer for businesses, governments and organizations. Combining security expertise and innovation, TrendAI™ leverages artificial intelligence to protect more than 500,000 companies and millions of individuals across AI, cloud, network and endpoint environments. Flagship platform TrendAI Vision One™ is an enterprise AI-driven cybersecurity solution that centralizes cyber risk and security operations, delivering protection across local, hybrid and multi-cloud environments. The platform is powered by threat intelligence and insights gathered worldwide, helping organizations defend against hundreds of millions of threats daily. With 6,000 security professionals in 75 countries, TrendAI™ enables organizations to stay ahead of threats and achieve proactive security outcomes across their attack surface, including critical environments such as AWS, Google, Microsoft and NVIDIA.
Read the full report here.