(519) 451-1793
For years, the advice for spotting a scam was simple. You'd look for spelling mistakes, check for strange sender addresses, and be wary of anything that feels "off", like the signature, or the font. That advice worked because most attacks were run by people, and people make mistakes.
That's no longer the case.
In July 2026, ransomware hit its highest monthly volume of the year. NCC Group (a British cybersecurity company) tracked 894 cases, which was a 22% jump from June, the month prior. That same month, researchers documented JADEPUFFER, the first known attack that runs entirely by an autonomous AI agent. It handled everything from the initial break-in to the ransom demand without the need for a human to direct it. Phishing and ransomware have been around for decades, but AI has completely changed the game with how fast and how cheap these attacks are to run.
The AI-Powered Attacks We're Seeing, and How We can Prevent Them
1. Phishing Emails That Look Perfect
AI can easily write flawless, natural-sounding emails in seconds. It can even personalize each email using information scraped from LinkedIn, your business website, or a previous data breach. Phishing emails created this way use correct contact names, and can even match the tone and formatting of past emails.
2. Deepfake Voices and Video
This is the attack that catches most people off guard. By using a snippet of audio from a voicemail, webinar or social media video, attackers can clone someone's voice. We're seeing it used in two ways: a "CEO" or "Owner" calling the finance team to push through an urgent payment, and similarly, an "employee" calling the help desk to get their password or MFA reset.
3. Vulnerabilities Exploited in Hours
AI tools can scan the internet for exposed systems and find weaknesses far faster than any human. IBM reported a 44% increase in attacks on public-facing applications, which they attribute to missing authentication controls, paired with AI-assisted vulnerability discovery. Old VPN tunnels, firewalls or remote desktop gateways are exactly what these tools are looking for, and many companies have vulnerabilities like these lingering without even realizing it.
4. Attacks on Your Own AI Tools
As businesses roll out Copilot, ChatGPT and other AI agents, attackers are targeting the AI itself. 'Prompt injection' hides instructions inside an email, document or web page that an AI assistant reads, tricking it into leaking data or taking actions the user never asked for.
How We Stop These New Attacks?
1. Phishing-resistant Sign-in
Passkeys make a stolen password useless to an attacker, and properly-implemented email security can catch impersonation attempts before the phishing emails reach your inbox.
2. Verification processes for money and access
Any request to move money, change banking details or reset credentials should be confirmed through a second channel. Second channels can take many forms, such as calling the person back on a number already on file, verbal (in-person) verification, or confirmation codes.
3. Fast patching and a smaller attack surface
Patches and updates should get approved and applied in days, rather than months. Furthermore, you should have a clear picture of what your business exposes to the internet. If something doesn't need to be public-facing, it shouldn't be.
4. 24/7 monitoring and response
If your security depends on someone checking alerts during business hours, an automated attack that starts at 2:00 AM on a Saturday has a full weekend head start. Automated systems should be in place to isolate threats and recover on their own, or at minimum, alert your team the moment something is accessed after hours.
Staying Ahead of AI-Driven Attacks with Keystone
The good news is that most of the controls that stop AI-powered attacks are the same ones insurers now expect every business to have in place, which we broke down in our "Cyber Insurance Requirements Are Changing: Is Your Business Still Insurable?" article. The difference now is speed. Gaps that used to take attackers weeks to find are being found in hours, and the businesses that close them first are the ones that stay off the list.
To find out where your business stands, visit keystonetech.ca, call 519-451-1793, or email info@keystonetech.ca.