For years, much of cyber security has been about detecting and responding to threats after something suspicious happens.
An unusual login is detected. A malicious file is blocked. A compromised account is flagged. Security teams investigate what happened and work to contain the damage.
That reactive approach remains essential. But what if cyber security could become more proactive — identifying vulnerabilities before attackers discover them?
That is the idea behind Microsoft MDASH, an AI-powered vulnerability discovery system that could offer a glimpse into the future of cyber security.
What Is Microsoft MDASH?
MDASH — short for Microsoft Security’s multi-model agentic scanning harness — uses AI agents to search software and code for potential security vulnerabilities.
Rather than relying on a single AI model, Microsoft has built a system that orchestrates more than 100 specialised AI agents, each designed to look for different types of weaknesses. The system can then validate potential findings and assess whether they represent genuine, exploitable vulnerabilities.
In simple terms, Microsoft is using AI to automate a task that security researchers have traditionally had to perform manually:
Find the security holes before criminals do.
And the early results are certainly interesting.
AI-Powered Vulnerability Detection At Scale
Microsoft says MDASH helped its security researchers identify 16 previously unknown vulnerabilities across the Windows networking and authentication stack.
Four of those vulnerabilities were classified as Critical remote code execution vulnerabilities, meaning that, under the right circumstances, an attacker could potentially execute malicious code on a vulnerable system remotely.
Microsoft has also reported strong results in controlled testing. On one private test driver containing 21 deliberately planted vulnerabilities, MDASH reportedly found all 21 with zero false positives. It also achieved an 88.45% score on the public CyberGym vulnerability benchmark.
That matters because finding a vulnerability is only useful if security teams can trust the result.
The False-Positive Problem
One of the challenges with automated security tools is noise.
A system might identify hundreds of potential security issues, but if many turn out to be harmless, security professionals have to spend valuable time investigating them.
Too many false positives can eventually lead to something almost as dangerous as missing a threat altogether: alert fatigue.
MDASH is designed to tackle this by using multiple specialised agents and additional validation processes to investigate potential vulnerabilities rather than simply flagging anything that looks suspicious.
Microsoft describes the system as being designed to discover, validate and prove the exploitability of vulnerabilities.
That shift — from “this might be a problem” to “this is a genuine vulnerability and here’s why” — could be particularly significant as software environments become increasingly complex.
Why AI Could Change Cyber Security
Modern organisations can have enormous amounts of software, cloud infrastructure, applications, devices and code to protect.
There are simply limits to how much human security researchers can examine manually.
AI agents can potentially work continuously and at a scale that would be extremely difficult for a human team to match.
That opens the door to a more proactive model of cyber security:
Instead of waiting for criminals to find weaknesses, organisations could use AI to search for them first.
The implications extend beyond Microsoft and Windows.
As AI-powered vulnerability discovery becomes more capable, we could see similar approaches being used to assess:
- Business applications
- Cloud infrastructure
- Web applications
- Network devices
- APIs
- Open-source software
- Custom-developed applications
- Configuration and identity systems
Microsoft has already expanded MDASH into preview availability for eligible organisations, suggesting that this is moving beyond a purely experimental research project.
Does This Mean AI Will Solve Cyber Security?
Absolutely not.
This is where it is important to keep the technology in perspective.
AI-powered vulnerability discovery is exciting, but it addresses only one part of the wider cyber security picture.
A business can have the world’s most sophisticated vulnerability-scanning technology and still be compromised because someone has:
- Reused a password
- Approved a fraudulent MFA request
- Clicked a convincing phishing link
- Failed to install a critical security update
- Given a user excessive access
- Lost an unprotected laptop
- Failed to maintain reliable backups
The National Cyber Security Centre (NCSC) continues to emphasise fundamentals such as securing important accounts with strong passwords and two-step verification, protecting devices and maintaining reliable backups.
In other words, AI does not make the basics obsolete.
If anything, it makes getting the basics right even more important.
The Cyber Security Fundamentals Still Matter
Keep Systems And Software Updated
Security vulnerabilities are regularly discovered in operating systems, applications and network equipment.
Applying security updates promptly reduces the window in which known vulnerabilities can be exploited.
Protect Accounts With MFA
A strong, unique password is important, but multi-factor authentication provides an additional layer of protection if a password is compromised.
The NCSC describes two-step verification as one of the most effective ways to protect online accounts where passwords are still being used.
Control Access
Users should have access to the systems and information they actually need — and no more.
Regularly reviewing user accounts, administrator privileges and former employees’ access can help reduce the potential impact of a compromised account.
Train Your People
Technology cannot prevent every successful phishing attack.
Your employees remain an important part of your security defences, which means they need to understand how to recognise suspicious emails, links, attachments and requests.
Maintain Reliable Backups
Backups remain one of the most important safeguards against ransomware, accidental deletion and other forms of data loss.
The NCSC recommends maintaining appropriate backups and ensuring that organisations know how to restore their data when needed. It also warns that simply relying on the built-in version history or recovery mechanisms of a cloud service may not be enough for critical information.
And perhaps most importantly: test your backups.
A backup that has never been successfully restored is not something you should rely on when your business is under pressure.
The Future Of Cyber Security Is Likely To Be Proactive
MDASH is interesting because it represents a broader shift in how cyber security could work.
For decades, defenders have largely reacted to what attackers are doing.
AI could increasingly allow security teams to get ahead of them.
Imagine security systems that continuously analyse software, configurations and infrastructure, identify potential weaknesses, determine which ones are genuinely exploitable and help security teams fix them before they become an incident.
That is a very different approach to security.
And it is particularly relevant as AI becomes more accessible to attackers too.
The same technology that can help defenders identify vulnerabilities could potentially help criminals discover and exploit them faster.
That could make the speed of vulnerability discovery — and, crucially, the speed at which organisations fix vulnerabilities — increasingly important.
AI Is Another Layer Of Defence — Not A Replacement For The Basics
The future of cyber security will almost certainly involve more AI.
We can expect AI to play a growing role in vulnerability management, threat detection, security monitoring, incident response and potentially even automated remediation.
But that doesn’t mean businesses should wait for the next breakthrough in AI security.
For most organisations, the biggest improvements still come from doing the fundamentals consistently well:
- Patch your systems.
- Protect your accounts.
- Use MFA.
- Control access.
- Train your people.
- Monitor your environment.
- Maintain reliable backups.
- Have a plan for when something goes wrong.
AI could become an incredibly powerful additional layer of protection.
But the strongest security strategy is unlikely to be AI instead of the basics.
It will be AI alongside strong security fundamentals, experienced people and well-managed technology.
The technology may be changing rapidly, but the goal remains the same: reduce opportunities for attackers, identify risks early and make it as difficult as possible for a cyber attack to become a serious business disruption.
Is Your Business Ready For The Next Generation Of Cyber Security?
You don’t need hundreds of AI agents working around the clock to start improving your cyber security.
You do need to understand where your biggest risks are — and make sure the fundamentals are properly covered.
If you’re not sure whether your business has the right protection in place, Aztek can help.
From managed IT support and cyber security monitoring to vulnerability management, Microsoft 365 security, penetration testing and incident response, we can help you build a security strategy that protects your business today while preparing you for what’s coming next.
Want to know where your cyber security could be improved? Get in touch.
