New Cyber Security Threats in the Era of AI

The AI-era threats defenders are seeing now

AI-powered phishing, deepfakes and voice cloning

The clumsy, typo-ridden phishing email is disappearing. Large language models produce fluent, well-formatted messages in any language, tailored to a target's role and company. More worrying is synthetic media: a few seconds of someone's voice from a webinar or voicemail is now enough to clone it convincingly, and video deepfakes have already been used to impersonate executives on calls and authorise fraudulent transfers. Defenders respond with out-of-band verification (confirming unusual requests through a second channel), stricter payment controls, and staff training that assumes a caller's voice and face can be faked.

Prompt injection and the OWASP LLM Top 10

When an organisation builds an application on top of a language model, it inherits an entirely new class of vulnerability. Prompt injection — hiding malicious instructions inside a web page, document or email that the model later reads — can trick an AI assistant into ignoring its rules, leaking data, or taking unintended actions. The OWASP Top 10 for LLM Applications catalogues these risks — prompt injection, insecure output handling, training-data poisoning, model denial of service, sensitive-information disclosure and more — and has quickly become essential reading for anyone securing AI products.

Adversarial machine learning and data poisoning

Machine-learning models can be attacked through their data and their inputs. In a data-poisoning attack, an adversary corrupts the training data so the finished model behaves badly or contains a hidden backdoor. In an adversarial-example attack, carefully crafted inputs — sometimes changes invisible to a human — cause a model to misclassify, slipping malware past an AI-based filter or fooling a fraud-detection system. Defending against these means thinking about the integrity of the data pipeline, not just the perimeter.

AI-accelerated malware and reconnaissance

AI tools help less-skilled attackers move faster: drafting malware variants, explaining exploit code, summarising stolen data and automating reconnaissance against a target. This lowers the barrier to entry and shortens the time between a vulnerability being disclosed and being exploited in the wild. It does not make attackers omnipotent — but it does mean defenders can no longer rely on attacks being slow or amateurish.

Model and data exfiltration

The models and datasets an organisation trains are valuable assets, and they leak. Model-extraction attacks query a deployed model enough times to reconstruct an approximate copy of it. Membership-inference and model-inversion attacks try to recover whether — or even what — specific records were in the training data, a serious privacy concern when models are trained on personal or proprietary information. Treating models and their training data as crown-jewel assets, with access logging and rate limiting, is now part of the defender's job.

Shadow AI inside the enterprise

Perhaps the most common AI risk in practice is mundane: employees pasting confidential code, contracts or customer data into public AI tools that the organisation never approved. This shadow AI quietly moves sensitive information outside the company's control. Governing it — through clear policy, approved tooling and data-loss controls — is one of the first things many security teams are now asked to do.

How defenders use AI in return

The same technology is a powerful ally for the blue team. Modern security operations centres (SOCs) use machine learning to spot anomalies in network traffic and user behaviour, to triage and summarise the flood of alerts, and to correlate signals that a human analyst would never have time to connect manually. AI copilots inside SOC platforms now draft incident summaries and suggest next steps, freeing analysts to focus on judgement and response. The point is not that AI replaces the analyst — it is that the analyst who can direct, verify and supervise these tools becomes far more effective than one who cannot.

That is the crux of the modern hiring brief. Employers are not looking for people who fear AI, nor people who blindly trust it. They want practitioners who understand how it fails and how to put guardrails around it.

The skills and roles employers are hiring for

Two role profiles dominate AI-era security recruitment, and they map neatly onto a learning pathway:

• SOC Analyst — the front line of defence. Monitors alerts, investigates incidents, hunts for threats and increasingly works alongside AI-augmented tooling. This is the classic entry point into a security career, and the foundations — networking, operating systems, security principles and incident response — are exactly what general cyber security training builds.
• AI Security Analyst — a newer, specialised profile. Secures the organisation's AI systems and uses AI to defend everything else: testing models for prompt injection and adversarial weakness, governing shadow AI, and applying frameworks like the OWASP LLM Top 10 and the NIST guidance on AI risk.

The grounding for both comes from solid fundamentals. The Advanced Certificate in Cyber Security builds the core competencies — system administration, network security, threat detection and hands-on incident response — that every SOC analyst needs, mapped to recognised industry certifications. From there, the Advanced Certificate in AI Security Analyst layers on the AI-specific knowledge: how models are attacked, how to test and harden them, and how to use AI tooling responsibly inside a SOC. Together they form a coherent route from beginner to a role that genuinely did not exist a few years ago. You can compare the full slate of programmes on the courses page.

What good AI-security training should give you

• Hands-on practice, not just theory — you should leave able to investigate a real alert and reason about a real model weakness.
• Coverage of established frameworks such as the OWASP LLM Top 10 and recognised AI risk-management guidance, so your knowledge is vendor-neutral and portable.
• A clear link to certifications and roles that employers recognise, so the credential translates into interviews.

Why study cyber security in Singapore

Singapore is one of Asia's most concentrated hubs for finance, technology and digital infrastructure, which makes it a serious market for security talent — and a practical place for an international student to build a career. As a registered Private Education Institution, Tertiary Infotech Academy focuses on industry-aligned, certification-mapped programmes rather than purely academic study, which suits learners who want to move into work. If you are weighing up a study-and-career pathway from overseas, the study in Singapore hub is the place to start your questions about courses, entry requirements and pathways.

What to do next

The AI era has not made cyber security harder to enter — it has made skilled, AI-literate defenders more valuable than ever. If you want to build toward a SOC or AI Security Analyst role, start with the fundamentals in the Advanced Certificate in Cyber Security, then specialise with the Advanced Certificate in AI Security Analyst. International students can begin an enquiry through the study in Singapore hub, or contact our team to book a course counselling call. Tell us where you are starting from, and we will map a pathway that fits.