Takeaways

  • AI has not created new cyber threats tactics, but rather has expedited the process for hackers. 

  • Attackers primarily infiltrate systems through Virtual Private Networks (VPNs) using leaked credentials, hijacked technical assistance tools and fake Captchas. 

  • Protecting the workplace against AI-driven attacks requires straightforward security measures, including limiting administrator rights, enforcing phishing-resistant MFS, disabling commands for non-IT users, establishing strict IT verification protocols and routinely auditing accounts and access portals. 

The modern dealership runs from everywhere—on mobile devices, through vendor portals and across cloud platforms. But as your workspace expands beyond the office floor, your digital identity (usernames, passwords and access credentials) becomes the primary entry point or perimeter for cyber threats.

“Everyone views their house as this private security sanctuary, assuming locked doors make it safe, but every door and window that you have is potential exposure,” Robert Russell, vice president of Response Operation Centers at Sedona Technologies, explains, comparing home vulnerability to digital network access. “If you leave those doors open, if those windows are open, the threats can walk right in.”

When an attacker compromises a password, they do not need to “break into” your network. Rather, they simply need to log in. 

“What we’d like for you guys to walk away with is knowing that your digital identity is the new perimeter. So these passwords that we all use from site to site to site, when that gets compromised, it turns into quite a big blast radius,” Art VanZuiden, Technical Account Manager at Sedona Technologies, says. 

Understanding AI Cyber Attacks

A common misconception about AI cyber attacks is that these are new or unfamiliar forms of attacks. In reality, AI acts as an accelerator for already existing threat methods. What used to require years of hacking skills or days of manual labor can be accomplished by low-skilled attackers in minutes using AI tools. 

“AI is being leveraged in such a way that it’s shrinking the amount of time that hackers used to have to spend in building things,” Russell says. 

To further show how attackers target dealerships, Robert Russell shared an analogy stemming from his previous career in law enforcement. When responding to a report of multiple car break-ins within the same parking lot, he expected to find physical evidence, such as smashed windows. Instead, he found unharmed vehicles that all had one thing in common. 

“The commonality among everyone that had their vehicle broken into is their doors were unlocked,” he said. 

The thieves had simply walked through the lot, jiggling every door handle. If a car was locked, they moved on. If it was unlocked, they took whatever was inside. 

Most cybercriminals work in the same opportunistic manner that the thieves did. They will scan the internet looking for easy targets, and AI allows the attackers to “jiggle thousands of door handles” across the world simultaneously. This results in a shrunken window of time that dealerships have to now detect and respond to the intrusion. 

Top 3 Digital Doorways Attackers Exploit

The Sedona Technologies’ Security Operations Center (SOC) team was able to highlight the three main entry points that are being used in cyber incidents. 

  1. Virtual Private Networks (VPNs) allow remote workers to access the company network. However within VPNs, any stolen or leaked passwords can lead the hacker to log in and appear as if they are a legitimate employee. Once they are logged in, attackers can get to the most valuable assets in a little as 7 minutes. 

  1. Authentic technical assistance tools, such as Quick Assist or Remote Monitoring software, are also frequently hacked into. Attackers are able to impersonate IT support via phone calls, emails or Microsoft Teams messages, conning employees into granting remote access. Since they are using legitimate software, the traditional antivirus software fails to flag it majority of the time.

  1. This final entry point is through Captcha, where users have to click the “I am not a robot” that often pop-ups on websites. Attackers will take the fact that Captchas become habitual, and create fake Captchas. Most of the time, users are not going to second guess it appearing. In a fake captcha, clicking the button secretly copies malicious code onto the user’s clipboard. The site will then prompt the user to press keyboard shortcuts, such as Window + R, Control + V then Enter, to verify themselves, causing the user to then run the malicious code on their own computer. 

Understanding Tier 1 vs. Tier 2 AI Threats 

Cyber threats involving AI can be divided into two main categories. Tier 1 focuses on current risks. These would be attackers using the off-the shelf AI to automate scanning, generate working malware scripts and write phishing emails that lack the traditional typos or phrasing to differentiate. 

“Hackers are utilizing AI to take advantage and exploit the vulnerabilities and the risks that are already present just in a much more expedient way,” Russell says. 

Tier 2 threats focus on emerging future risks. These can be autonomous AI agents capable of executing complex, multi-step attacks at machine speed. 

The Hugging Face, an AI company, breach is a good example of differentiating between Tier 1 and Tier 2. In this breach, an AI agent that was under testing by OpenAI and “broke out of the sandbox.” It used leaked credentials and flaws in Hugging Face’s own data pipeline. From there, it climbed from a single admin account into multiple different clusters within 13 hours and rolled itself into the internal network with a stolen VPN key. It was cut off after almost 18,000 actions. In the post-incident technical timeline, they said one stolen system credential became  allowed full administrator control across all system groups.

“What’s really important to take away from that is this is showing that AI uses this Tier 2 problem that I have of using things like stolen credentials, using things that are open and available in the data pipeline already, using those stolen credentials to log in,” Russel says. “And then it becomes tier two level at the start of taking 18,000 different actions.” 

While knowing the difference between the two tiers is important, Russell suggests prioritizing the focus on Tier 1, as it focuses on current threats. 

Practical, Actionable Guidance 

Defending against AI-driven access abuses does not require extensive knowledge or complex technology. The most important defense is to implement these security housekeeping items into dealerships:

  • Limit administrator rights to only what is necessary for a job function. Everyday tasks such as searching the web or checking emails should not be done using an administrative account. 

  • Require phishing-resistant Multi-Factor Authentication (MFA) on all VPN portals. Firewall management screens should never be directly exposed to the public internet. 

  • Disable the Windows Run command box for non-IT users via Group Policy settings to eliminate fake Captcha attacks. 

  • Establish strict verification protocols for IT requests. Employees should be trained to hang up and call IT back using a verified internal phone number whenever an unknown caller requests access or passwords. 

  • Perform routine inventory checks to remove expired employee accounts, close inactive vendor portals and delete unapproved remote access software. 

Watch the full webinar, “The New Perimeter Problem: AI, Access Abuse, and the Modern Threat Landscape”, here. 

Frequently Asked Questions

Q: How does AI accelerate existing cyber attack methods against businesses? 

A:  AI acts as an accelerator for existing threat methods, rather than creating entirely new attack tactics? Cyber criminals use AI to automate system scannigng, build working malicious scripts and write polished phishing emails in minutes rather than days, drastically reducing the time organizations have to respond. 

Q: What are the top digital entry points hackers exploit in cyber attacks? 

A: The three primary entry points exploited by attackers are: Virtual Private Networks (VPNs), Technical assistance tools and fake Captchas.

Q: What is the difference between Tier 1 and Tier 2 AI threats?

A: Tier 1 focuses on current risks, where attackers use available AI tools to automate scanning, write malware and eliminate typos from phishing emails. Tier 2 focuses on emerging future risks, where autonomous agents capable of executing complex attacks at machine speed across internal networks.