Is this course right for me?

Target Audience: Academic, Research, and KE staff

We invite you to the eleventh workshop in our ML/AI series, which brings together colleagues from across the University to explore opportunities for collaboration.

This time, we have two presenters from Computer and Information Sciences, Faculty of Science, speaking on the implications of AI for cybersecurity, a timely topic with consequences for research and everyday digital security. The talks examine the security, privacy and safety challenges of agentic AI and how generative AI is being adopted within cybercrime ecosystems.

Programme

3:00 - 3:05  Welcome - Professor Sergey Kitaev

3:05 - 3:25  Why Agentic AI strains Security, Privacy and Safety - Dr Jide Edu (CIS)

3:25 - 3:45  Exploring the adoption and innovation of GenAI tools, coding assistants, and agents within cybercrime ecosystems - Dr Daniel Thomas (CIS)

3:45 - 4:00  Questions and discussion

Registration

Please register by 12 noon on Wednesday 25 November 2026.

For workshop enquiries, please contact Professor Sergey Kitaev (sergey.kitaev@strath.ac.uk), Associate Dean (Research)

Abstracts

Why Agentic AI strains Security, Privacy, and Safety

Dr Jide Edu (CIS)

AI systems are increasingly acting on behalf of users rather than merely responding to them. Agentic AI can plan, use tools, delegate tasks, and retain memory across services, so a harmful disclosure, purchase, or tool call may occur before a user has a chance to review it.

We argue that this shift strains four assumptions underlying existing safeguards: that agent behaviour is bounded, attributable, stable, and intervenable.

Informed by a horizon-scanning workshop, we organise the resulting research agenda around four questions: Who is responsible? What did the user authorise? Whose interests is the agent serving? And what happens when it fails? Together, these questions connect security, privacy, and safety to safeguards that track agentic action as control and behaviour change over time.

Exploring the adoption and innovation of GenAI tools, coding assistants, and agents within cybercrime ecosystems

Dr Daniel Thomas (CIS)

Existential risk scenarios relating to Generative Artificial Intelligence often involve advanced systems or agentic models breaking loose and using hacking tools to gain control over critical infrastructure. In this paper, we argue that the real threats posed by generative AI for cybercrime are rather different.

We analyse early empirical data from a variety of large-scale digital sources from the cybercrime underground, and find that the reality is prosaic -- AI is seeing some early adoption in existing large-scale, low-profit passive income schemes and forms of fraud but there is little evidence that it is so far giving rise to widespread disruption in technical aspects of cybercrime.

The role of jailbroken LLMs (Dark AI) as hacking instructors is overstated, given the prominence of subculture and social learning in initiation -- new users value the social connections and community identity involved in learning hacking and cybercrime skills as much as the knowledge itself. Disruptive adoption is mainly in the domain of logistics and automation, especially in bot farming and the production of approach emails and social frauds.

Our initial results, therefore, suggest an intensification and scaling of the dynamics of cybercrime-as-a-service rather than a rupture.

Prerequisites