top of page

AI isn’t smarter when it comes to cyber attacks. Just a lot faster.

AI isn't making cyber attacks smarter – it's making them faster. A lot faster. In this piece, Baidam Pre-Sales Architect Michael Choo breaks down what that actually means for your defences, and where to start.

We're as guilty as anyone of saying AI enables cybercriminals to make their attacks smarter. Because the truth is, AI just makes them faster.


While AI increases risks tremendously – and AI-powered attacks are skyrocketing, reducing attack timelines from weeks to minutes to a record 27 seconds – it's not doing anything new or clever in how it attacks.


AI-powered attacks don't use novel techniques to exploit victims' environments. They exploit the same vulnerabilities, using the same tactics and techniques that human threat actors have used for years.


So, what’s the issue?


AI vs human threat actors


The problem is that unlike humans, AI works 24/7.


AI has no need to sleep, eat, or spend time at the coffee machine. With the processing power available to it, it can discover and exploit vulnerabilities far faster than any human team could. It may look smarter – but AI's ability to process petabytes of data in seconds is what makes the difference, and what makes it so dangerous.


While its speed of attack may seem insurmountable, remember it can only ingest, analyse, and act on what we, as humans, provide. It relies on what it's been trained to find, using proven tactics and techniques programmed into it. It doesn't (at this stage anyway) create new types of attacks.


In short, since AI-enabled attacks exploit existing weaknesses, the usual best-practice approach to securing the attack surface still applies.


But that comes with a big caveat.


Here's the catch


Knowing that AI isn't smarter, just faster, may be comforting. But it doesn't mean cyber security can be managed the same way it could a few years ago. It can't.


The difference is that we now must fight fire with fire. And AI speed with AI speed.


AI has a clear advantage over humans in storing and processing information. Our biological limitations mean we simply can't compete with the rate at which AI attack agents can identify and exploit vulnerabilities.


Yet our ability to detect, respond, and act with equal or greater speed is now critical – and that means using cyber security tools that are also AI-enabled.


But before we get to tools, we need to talk about what you're actually defending.


Know your exposure – end to end


Security thinking is shifting from talking about vulnerabilities to talking about exposure. The difference matters.


A vulnerability is a weakness in a system. Exposure is your understanding of all the weaknesses across your entire environment – from your internet-facing edge, through your internal network, your applications and servers, all the way down to your data and identity.


Most organisations have gaps in that picture.


And exposure management isn't just about discovery. It's about prioritisation in context. A critical vulnerability being actively exploited in Europe but with no recorded activity in Asia Pacific is a very different risk from one currently circulating in your industry vertical. Threat intelligence, business context, and regional relevance all factor into how you triage and respond.


The faster attackers can move, the more important it is to know where your weak points are before they do.


Identity is at the heart of it


When you strip a cyber attack back to its core, it's almost always about data. And the most reliable path to data is through identity.


Organisations have invested heavily in network security, endpoint protection, perimeter defence, and, of course, identity security. The problem is that much of that investment was built for a different era - one where identity-based attacks were slow and predictable, and the perimeter still meant something.


Identity is now a primary attack surface. Today’s adversaries move through it fast, using MFA bypass techniques, stolen session cookies, and credential stuffing to get inside organisations in minutes. CrowdStrike recorded a 50% increase in Access Broker Activity and a 442% increase in voice phishing (vishing): techniques aimed squarely at identity, not endpoints or perimeters.


Detection and response for identity-based events is still catching up. Meanwhile, 90% of Fortune 1000 organisations still run Active Directory - a platform built well before anyone was doing this to it. Right now, that gap is where most intrusions start.


And here's where it gets more interesting: AI agents are identities too. Every agent running in your environment – every automated workflow, every AI-powered tool – is a non-human identity with its own access and permissions. That's a rapidly growing attack surface that most organisations haven't fully accounted for yet.


Fight AI with AI


So, how do you defend against an attacker who moves faster than any human team can respond?


You use AI.


AI-powered defence isn't optional. It's the new cyber security baseline. You can't match the speed of a machine with manual processes, no matter how capable your team is. Don't bring a knife to a gunfight – bring a radar-guided, imaging infrared missile.


In practice, AI defence breaks down into three areas:


  • Workforce – protecting your people,      their devices, their email, and the AI tools embedded in their day-to-day      work, including copilots and AI assistants built into existing platforms

  • Agentic – securing the AI agents      your organisation uses to automate tasks, including monitoring for prompt      injection, data leakage, and unexpected behaviour

  • Applications – protecting the AI      capabilities built into the applications your teams use or your developers      are building


Each area requires its own approach, and there's genuine overlap between them – which means the tools you choose need to work well together.


Cyber security is a team sport


No single vendor covers all three areas effectively. AI defence is its own category now, and we believe the right answer is rarely one platform – it's a considered mix of solutions, each playing to its strengths.


The challenge is that every vendor currently sounds like every other vendor. The marketing is almost indistinguishable. What actually differentiates them is their heritage , where they came from and what problem they were originally built to solve. An identity-native solution approaches security very differently from a platform player that has added identity capabilities over time. A specialist agentic AI security tool has different strengths from a broad endpoint platform expanding into agentic security.


Getting this right requires a genuinely vendor agnostic view of what each solution actually does well – and where the gaps are.


Australia is also a bit different. Unlike some other markets, Australian enterprises tend to consolidate rather than accumulate. They typically prefer to extend existing investments where possible and introduce specialist solutions only when the existing stack genuinely can't do the job. A solution that meets 70% of requirements within an established ecosystem will often win over a best-of-breed standalone that brings new integration overhead. That context matters when you're evaluating your AI defence options.


Where to start


If you're not sure where to begin, here’s where to focus first:


  1. Audit your AI footprint – identify      every AI tool in use across your organisation, including shadow AI and      tools your teams have adopted informally

  2. Assess your identity security – not      just who has access, but how AI agents and non-human identities are      governed, monitored, and controlled

  3. Review your detection and response capability – can you detect and respond to events at the speed AI      operates, including identity-based ones?

  4. Map your exposure end to end –      understand where you're most vulnerable before an attacker finds out for      you

  5. Evaluate your AI defence stack –      make sure the tools you're using have a genuine AI capability, not just      AI-flavoured marketing


Fighting fire with fire starts here


The organisations handling this best aren't necessarily those with the most tools.


They're the ones who understand their own environment, have invested in the right defences across their key risk areas, and have AI-powered monitoring and response capabilities that keep pace with the threat.


If you'd like to talk through what that looks like for your organisation, why not have a yarn with one of our experts?

bottom of page