
Artificial intelligence has become the most powerful tool in cybersecurity, for defenders and attackers alike. In 2026, that dual role is reshaping not only how attacks happen, but who is hired to stop them, what those people need to know, and how they learn it. This article follows that chain from start to finish: the threats, the jobs they’ve created, and the paths that help professionals stay ahead.

Part 1: The threat - how AI changed cyberattacks in 2026
The pace of change has been striking. CrowdStrike reported that AI-enabled cyberattacks nearly doubled over the past year, as threat actors used machine learning and large language models to sharpen their techniques.
Faster, more convincing social engineering
Phishing emails, voice cloning and deepfake video are now cheap to produce at scale and far harder to spot. Google's threat researchers note that the old warning signs, such as poor grammar or awkward phrasing, are disappearing, as attackers use AI to write personalised, culturally fluent messages and even hold believable multi-message conversations before striking (Google Threat Intelligence Group, February 2026).
Supply chains and AI tools as entry points
IBM's X-Force Threat Intelligence Index 2026 found that major supply chain and third-party breaches have quadrupled over five years, and recorded a 44% year-on-year rise in attacks on public-facing applications. AI tools themselves are now targets: X-Force found more than 300,000 ChatGPT credentials for sale on the dark web in 2025.
From experiment to autonomy: one year, three milestones
At the start of 2026, many experts believed truly AI-powered attacks were still some way off. Google's own reporting shows how quickly that changed:
February 2026: Google found attackers experimenting with AI-enabled malware, but concluded these experiments had not yet produced any revolutionary shift in the threat landscape (GTIG).
September 2026: AI agents were taking on larger parts of attacks, such as vulnerability scanning and credential harvesting, but Google had not yet seen fully autonomous attacks in the wild (Help Net Security).
16 November 2026: Google declared that "the era of autonomous attacks has arrived" (GTIG).
At the centre of the November report is a tool researchers named "exploit_agent", found on an exposed server. It compares software code before and after a security patch, works out which flaw was fixed, builds an exploit and launches it, with no human involved (The Decoder). In other words, attackers can now turn a security update into an attack before many organisations have installed it.
Google expects this to escalate quickly, with simple autonomous attacks likely within one to three months and more sophisticated ones within three to six (Crypto Briefing). Its message to defenders is blunt: machine-speed threats cannot be countered at human speed alone. There is some reassurance, though. For now, these automated attacks trade stealth for speed, leaving forensic traces that make them easier to detect.
This also explains why AI-powered malware remained rare for much of the year. The breakthrough did not come from AI hidden inside malicious code, but from AI running the attack process itself, from finding the flaw to launching the exploit.
AI models are now targets too: model extraction attacks
Attackers are not only using AI, they are also stealing it. In a model extraction attack, also called a distillation attack, someone with ordinary access to an AI model sends it huge numbers of carefully chosen questions, collects the answers, and uses them to train a copycat model. It is a form of intellectual property theft that lets the copier skip years of costly development (GTIG).
Google reported one such campaign using more than 100,000 prompts in an attempt to replicate its Gemini model's reasoning abilities. Google stresses that this kind of attack is not a direct risk to everyday users. The risk falls on the organisations that build and offer AI models, from AI labs to a company that has trained its own model on financial or medical data.

Why this matters for careers: when the attack surface shifts to AI models, agents and automated pipelines, organisations need people who understand both security and AI, and who can respond at machine speed. That is exactly where the job market has moved.
Part 2: The response – how cybersecurity jobs have changed in five years
In 2021, a typical security team was built around well-established roles: SOC analysts monitoring alerts, penetration testers probing systems, security engineers hardening infrastructure, and compliance analysts managing audits. Those roles still exist, but in 2026 many have been rewritten around AI.
The market is shrinking in some places and surging in others. CyberSN recorded 235,786 US cybersecurity job postings from January to September 2026, down 26.4% on the same period in 2025. Yet demand for AI Security Engineers rose 268.3% and for Cyber Data Scientists 203.5%. Globally, cybersecurity job demand in G7 economies grew 9.5% in the six months to March 2026, with senior-titled postings up 65%.
Entirely new categories are emerging. The SANS 2026 Cybersecurity Workforce Survey found that demand for specialists in new roles more than doubled in a year, with employers adding AI security engineers, AI governance analysts and AI/ML security specialists. Regulation is a driver too: frameworks such as NIS2 and DORA are shifting hiring priorities, which matters for anyone building a career in Europe.

Here is how the shift looks role by role, and where each skill set is covered in the MSc in Applicable AI & Cybersecurity:
2021 role | 2026 role | What changed | Related MSc modules |
|---|---|---|---|
SOC Analyst | AI-Augmented Detection & Response Engineer | Alert triage is increasingly automated; analysts supervise AI tools, verify their output and respond at machine speed | AI-driven Threat Detection & Incident Response; Applied Machine Learning for Security Analytics |
Penetration Tester | AI Red Teamer / LLM Security Specialist | Testing now covers models, prompts and agents, not just networks and apps | AI Security I: Adversarial ML, LLM Security & Red Teaming |
Security Engineer | AI / ML Security Engineer | Protecting the AI pipeline itself, including training data, models and defences against model extraction | Secure Cloud Architectures & DevSecOps; MLOps/LLMOps & Reliable AI Operations |
GRC / Compliance Analyst | AI Risk & Governance Specialist | The EU AI Act and data protection rules extend compliance to AI systems | AI Governance, Ethics & Compliance; Privacy-Preserving Machine Learning |
Security Data Analyst | Cyber Data Scientist | Defence relies on modelling large volumes of security data | Applied Machine Learning for Security Analytics; Data Engineering for AI Systems |
Cloud Security Architect | Secure AI Systems Architect | Architectures must account for autonomous agents and identity-first security | Agentic AI & Secure Automation; Cryptography, Zero Trust & Identity |
One pattern deserves attention. Senior cybersecurity roles grew 65% while junior roles grew only 5.9%, revealing an experience gap around working with AI agents and security fundamentals. Employers increasingly want depth, not just familiarity.
Looking ahead: with autonomous attacks now a reality, human roles are likely to move further towards oversight, governance and judgement: deciding what AI should do, checking what it did, and taking responsibility for the outcome. That raises the question of how professionals build that depth.
Part 3: The path – how to keep up
New roles call for new proof of skill. Certification bodies have responded quickly, and two credentials stand out in 2026.
1. CompTIA SecAI+
Launched in February 2026, SecAI+ focuses on securing AI systems and using AI in everyday cybersecurity work. It covers AI in incident response, security analytics, threat intelligence and penetration testing, as well as AI-supported compliance and risk management, and is designed to sit alongside established credentials such as Security+, CySA+ and PenTest+. It suits practitioners: CompTIA recommends 3–4 years of IT experience and at least 2 years of hands-on cybersecurity work.
2. ISACA Advanced in AI Security Management (AAISM)
Where SecAI+ is hands-on, AAISM is aimed at security leaders. It is positioned as the first AI-centric security management certification, and candidates must already hold a CISM or CISSP. Its curriculum spans AI governance and programme management, AI risk management, and AI technologies and controls.
Both are valuable, but both assume an existing foundation, and both validate knowledge rather than build it from the ground up. For anyone who wants to design, secure and govern AI systems end to end, a Master's degree offers that depth.
3. MSc in Applicable AI & Cybersecurity
The MSc in Applicable AI & Cybersecurity at Steinbeis Next was built around the shift described in this article. Students learn to design and operate AI systems in secure cloud environments using DevSecOps practices, and to identify and reduce AI security risks through methods such as adversarial machine learning and model red teaming, the same skills needed to defend against threats like model extraction. The programme also covers integrating EU AI regulation and data protection into technology development.
The modules respond directly to this year's headlines. AI-driven Threat Detection & Incident Response prepares students to defend at machine speed, while Agentic AI & Secure Automation explores the same kind of AI agents now being used in autonomous attacks, from the defender's side.
The career outcomes map onto 2026's fastest-growing roles: AI Security Engineer, AI Systems Architect, Cybersecurity Analyst, AI Risk & Governance Specialist, Machine Learning Security Engineer and Security Data Analyst. An industry or transfer project in the final semester gives students the chance to apply their skills to real organisational challenges before graduating.
At a glance: four semesters, full time in Berlin, taught in English, 120 ECTS, with intakes in October, January, April and July.
Certifications and a Master's are not competing choices. Many professionals use a degree to build deep, broad foundations, then add certifications like SecAI+ or AAISM to signal specialist skills as their career develops.
From threat to opportunity
AI has changed what attacks look like, and as of this week, it can run them on its own. That has changed which roles organisations hire for, and in turn, what professionals need to learn. The people who will thrive are those who understand both sides: how AI can be turned against systems, and how to build systems that hold up.
If you are ready to be part of that next generation, explore the MSc in Applicable AI & Cybersecurity or book a call with our study advisors.
Sources
IBM: Cyberthreats in 2026, X-Force and industry experts weigh in
Infosecurity Magazine: AI-powered cyberattacks up significantly, warns CrowdStrike
Checkout what our students have to say about studying with us:




