Cybersecurity Trends in 2025

Threat Agents and Human Behaviour

The next session in the trilogy of webinars looking at cybersecurity trends and challenges during the course of 2025 focussed on threat agents and human behaviour.

Summary

The webinar on cybersecurity trends and human behaviour featured a panel discussion on AI’s role in both cyber threats and defences. Participants explored challenges around changes in behaviours, awareness campaigns and the integration of AI capabilities in organisations, as well as discussing the evolving nature of cyber threats and the importance of digital autonomy. The panel concluded by addressing future cybersecurity trends, including the need for simplified user experiences, adaptive malware and the psychological burden of constant verification, while emphasising the importance of balancing AI security investments with preparations for emerging threats.

Background

Despite additional guardrails in technologies, best practices and regulatory frameworks, human error and human behaviour continue to be among the top causes of cybersecurity events. These trends have been amplified recently by AI-enabled social engineering. A panel of experts were invited to discuss the state of this field in 2025, which included increased vulnerability from human factor-enabled events, the impact of the fragmented regulatory space, progress in behaviour-focused cybersecurity, the evolution of insider threats and strategies for protection from threat agents that focus on human factor related attacks.

Each participant was asked to give their views on the most important topics, challenges and lessons learned in human factor related topics in cybersecurity in 2025. The initial questions asked what in 2025 were the most important broad trends and challenges in human factors in cybersecurity (and whether they will continue into 2026) as well as both the most negative and most positive developments during the year. The next set of questions sought to understand how new technologies, including AI, influenced these aspects of cybersecurity and what the determining factors for the most impactful trends were: technology issues, geopolitical developments, lack of training, inadequate organizational structures or anything else.

It was noticed a long time ago that attack mitigation strategies depend heavily on the threat agent performing the attack. So, another question was whether the engagement of different threat agents, from disgruntled employees to nation states, remained relatively static or whether it evolved over the last several years. And, if it did evolve, what defined the evolution and what the most dangerous trends with regard to the cybersecurity threat agents are today: for example, whether economic considerations still prevail in social engineering attacks.

It has been observed that the usability of user and organisational protection technologies need to be adequately addressed to ensure they are used as designed. This raises the questions as to how organisations are adapting, what are considered the most important trends in cybersecurity that organisations face today and whether the main challenges are impacted by technology, process, regulatory requirements or other factors.

The final set of questions focussed on the future of these aspects of cybersecurity and what areas, be they process, training, technology, organisational structures, etc, will have the most positive impact on human aspects of cybersecurity as well as what trends will be the most negative.

A lot to consider ... see how far we got!

Speakers

The distinguished panel comprised:

• Koen Maris, Head of Cybersecurity Advisory, Atos (Belgium)

• Massimo Felici, Security Consulting Senior Manager, Accenture (Belgium)

• Mike Loginov, Co-founder Executive, The Cybersecurity & AI Governance Initiative (CAGI)

• Wolter Pieters, Professor of Work, Organisations and Digital Technology, Radboud University

The session was moderated by TDL strategic advisor, Claire Vishik.

AI Cybersecurity Learning Challenges

The panel were invited to share their insights on the human aspect of cybersecurity, including the role of AI in both attacks and defences. The discussion started by highlighting a shift from awareness to behavioural change, emphasising the use of micro-learning in this context.

The need for continuous learning and awareness campaigns to address evolving AI-driven cyber threats was further emphasised, as traditional periodic awareness efforts are no longer sufficient. It was suggested that experiencing problems firsthand is a more effective learning method than passive information dissemination. It was pointed out that the rapid adoption of AI technologies, which creates both business opportunities and new security threats, while simultaneously exacerbating existing talent shortages in both AI and cybersecurity skills, leads to a widening gap between technological advancement and security solutions.

AI Cybersecurity: Challenges and Solutions

While AI offers significant opportunities in cybersecurity, it was highlighted that organisations need to focus on personalised solutions and cultural shifts to effectively integrate AI without eroding human expertise. Further mentioned as important was training workforces to leverage AI capabilities while maintaining vigilance against new threats. The discussion moved to the evolution of cyber attacks, noting that attackers are now using AI to engineer mistakes, making it difficult for human intuition to detect threats, which raises concerns about AI’s ability to manipulate human behaviour using psychometric data.

Enhancing Cybersecurity Through Human Trust

It was argued that traditional training approaches are insufficient for modern cybersecurity threats, emphasising the need for technology to protect humans as adversaries become more sophisticated. It was suggested that, with the rapid pace of AI-driven attacks, trust in technology must be developed, with questions around controls and governance. Could standardising human behaviour improve technology’s effectiveness? To which the response was that while humans are pattern-based, legal constraints limit potential actions.

AI Standardisation and Governance Challenges

The discussion moved on to the challenges of standardising human behaviour in technology and the need for international governance frameworks for AI deployment. The importance of control mechanisms and testing standards for AI capabilities was emphasised, while the difficulties of achieving international standards due to geopolitical and military considerations were highlighted. The ensuing conversation discussed behaviour change techniques and the shift from awareness campaigns to more comprehensive approaches in organisational structures.

Collaborative Cybersecurity for Behaviour Change

The need to shift from individual cybersecurity awareness to a more social and collaborative approach was mentioned, with an emphasis on the importance of embedding security practices within the work design to reduce overload and promote behaviour change. The role of human factors in incident responses was highlighted, suggesting that understanding collaboration within teams and the interaction between humans and AI could improve cybersecurity effectiveness. It was also noted that, while AI and rapid attack evolution pose challenges, traditional methods like phishing exercises, while useful, may not be entirely effective in preventing breaches.

Enhancing Cybersecurity Behaviour Training

The discussion focused on the effectiveness of behaviour changes and awareness efforts in cybersecurity, emphasising that reducing risk should be the primary goal rather than awareness itself. It was noted that phishing training statistics remain static despite repeated efforts, highlighting weaknesses in current approaches. While it was agreed that threat agents haven’t fundamentally changed, the volume and speed of attacks have increased, suggesting a need for more effective training methods that move away from simple compliance checks and towards hands-on, experience-based learning.

Evolving Cyber Threats and Vulnerabilities

The discussion focused on evolving cyber threats and organisational vulnerabilities. Learning from failure was considered important which highlighted a decline in common sense due to over-reliance on technology. Changes in the cyber threat landscape were discussed, particularly the increased intensity and volume of cyber threats, as well as the growing exposure of organisations to cyber-physical threats. It was pointed out that there is a shift in how organisations are exposed to cyber threats, particularly due to changes in global supply chains and geopolitical events.

Digital Autonomy in Cybersecurity

The discussion shifted to recent changes in cybersecurity, noting an increase in low-tech crimes like fake police officer scams in the Netherlands due to improved digital defences, alongside more sophisticated “cyborg” attacks using AI. This highlighted the growing importance of digital autonomy, which requires organisations to rely less on traditional security measures and potentially sacrifice some individual autonomy. It was agreed that digital autonomy in cybersecurity is complex, as increased autonomy can lead to reduced protection and access to obvious security measures.

AI Security Challenges and Sovereignty

The discussion turned to cybersecurity challenges, particularly the intersection of physical security and AI threats. focussing on the enduring effectiveness of physical security breaches and the new challenges posed by AI agents, which can make unauthorised data exfiltration difficult to detect. While data sovereignty is a more manageable issue, achieving AI sovereignty is complex due to the difficulty of controlling AI’s access to and use of data across borders. It was suggested that, despite the opportunities AI offers, many organisations lack AI capabilities, and sovereignty remains an important consideration for organisations seeking to protect their data.

AI Security Governance Framework

The panel discussed security governance for AI, focusing on four key actions: developing security elements in AI data, building AI resilience systems, embedding AI capabilities in organisations and addressing digital sovereignty. The importance of organisational decision-making in cybersecurity was stressed, particularly in the context of digital autonomy and human factors. The panel concluded by discussing future developments in cybersecurity, mentioning the need for further discussion on the topic.

Future Cybersecurity Trends and AI

The panel discussed future cybersecurity trends, emphasising the need for simplified user experiences and anticipatory governance, particularly regarding AI. Among the predictions were the rise of adaptive malware, the death of the perimeter and increased focus on zero trust and biometric authentication. Concerns were raised about the psychological burden of constant verification and the need for holistic approaches to responsible working, with additional warnings about widening gaps in AI security capabilities and the importance of preparing for post-quantum encryption. The panel agreed that while AI presents both opportunities and challenges, organisations must balance their investments in AI security solutions with preparations for emerging threats.