Artificial intelligence is becoming a common tool in cybersecurity, helping teams process large data sets, automate routine tasks and accelerate investigations. But an industry commentator argues that excessive reliance on AI could weaken some of the human capabilities security organizations need during unfamiliar or high-pressure incidents.
The concern centers on the difference between using AI as an aid and treating it as a substitute for professional judgment. In the article, the author describes an interaction in which a prospective event speaker asked for ready-made presentation material rather than seeking advice on how to develop it. The exchange prompted a broader question: whether professionals are beginning to approach colleagues and complex work in the same transactional way they use generative AI prompts.
Skills at risk
The author identifies several areas that may suffer if AI is used too broadly in security work:
- Creativity: Analysts and researchers often spot unusual patterns or develop unconventional approaches that automated systems may not prioritize.
- Critical thinking: Novel incidents may lack historical examples or established response playbooks, requiring teams to reason through incomplete evidence.
- Communication: Security teams depend on concise, accurate messaging during incidents. Overuse of generated text could contribute to vague or repetitive communications.
- Human connection: Humor, empathy and sincerity can help security leaders deliver difficult news, gain cooperation and build trust across an organization.
- Distinctive messaging: AI-generated marketing and conference content can produce similar language across vendors, making it harder for organizations to communicate a clear point of view.
Setting boundaries
The argument is not that AI lacks cybersecurity value. Automation can improve productivity and help defenders manage growing volumes of alerts and information, while attackers are also using AI to increase the pace of their operations. Instead, the author calls for deliberate boundaries around tasks that should remain human-led.
Examples may include forming incident hypotheses, making high-stakes risk decisions, developing original security strategies and communicating with affected stakeholders. Organizations can still use AI for support, such as summarizing information or handling repetitive workflows, while requiring analysts and leaders to validate outputs and retain responsibility for conclusions.
The central message is that security teams should adopt AI without allowing convenience to erode the curiosity, judgment and communication skills needed when technology cannot provide an established answer.
