Artificial intelligence is making business life a whole lot easier. With AI, we can quickly build websites, automate tasks, develop apps, and boost productivity all without needing to be a tech whiz.
But there’s a question that’s not being asked enough: what happens when those AI tools you’re using introduce security risks you don’t even know exist?
A recent incident involving a Claude Code extension highlights just how easily a trusted development environment can become a backdoor for hackers.
And while this particular attack targeted devs, the lessons learned extend to any business that’s starting to use AI.
A Real-Life Example: How A Rogue Claude Code Extension Blew Open 5,000 Devs’ Computers
On October 29, 2026, a malicious version of Anthropic’s Claude Code extension popped up in the Cursor extension marketplace.
Cursor’s an AI-powered code editor that helps devs build software using AI. Claude Code’s an AI coding assistant that can help write, debug and modify code.
According to a report on October 30, the compromised extension was available for around 11 hours, and potentially exposed up to 5,000 developers.
The malicious extension was designed to slip in an SSH backdoor, which would have given the attackers full remote access to the affected computers.
Once installed, the extension would have collected all sorts of info about the user’s computer and dev environment, added an attacker-controlled SSH key to let them get in remotely, and sent system info and connection status to an attacker’s server.
Luckily, researchers found coding errors that likely stopped the bad guys from getting in on many affected systems. But the malware could still establish a foothold, and send back all sorts of info about the affected dev environment.
Cursor yanked the malicious extension, shut down its marketplace for a bit, and took steps to contain the mess.
The important thing to take away here is that legitimate, widely-used tech can introduce big risks when a compromised extension or integration gets inserted into the mix.
Why This Should Be A Wake-Up Call For Business Owners Using AI
You don’t need to know how to code to get caught out by the risks associated with AI tools.
More and more businesses are using AI to:
- Build and update websites.
- Develop custom apps and customer portals.
- Automate admin tasks.
- Connect CRM systems with other apps.
- Generate reports and analysis for business data.
- Manage customer comms.
- Build AI assistants that interact with internal business systems.
These tools can save time and cut costs, but many need access to files, apps, accounts or sensitive business info.
And that’s where the risks start.
Take a business owner who uses an AI coding assistant to build a new website. The AI generates the code, recommends extensions, and helps connect the website to a database or payment platform.
But who’s checking to see if the extensions are legit? Who’s reviewing the code for security holes? And who’s making sure the AI tool isn’t accessing data it shouldn’t be?
AI makes building tech easier, but that doesn’t mean it makes the tech secure.
5 Hidden Cybersecurity Risks Of AI Tools
1. Malicious Extensions and Third-Party Integrations: When What You Add Can Spy On You
Many AI platforms rely on extensions, plugins and integrations to do more. But these additions can connect AI tools to dev environments, cloud platforms, databases and business apps.
The problem is, they can also provide a backdoor for hackers.
The Claude Code incident is an example of a software supply-chain attack, where the bad guys slip in malware through a component you think is trustworthy.
A business might carefully vet an AI platform, but overlook the security of the extensions connected to it.
What businesses should do: Have a process in place to approve AI tools, extensions and integrations before you install them. Check the publisher, permissions, distribution source and security history before you go ahead.
2. AI-Generated Sites and Apps Can Have Big Security Holes Too
AI coding assistants can generate pretty good-looking sites and apps in no time.
But that generated code can have some serious security issues if it’s not reviewed and tested properly.
Some potential problems include:
- Insecure logins and authentication.
- Bad handling of customer info.
- Exposed API keys or database credentials.
- Vulnerable third-party dependencies.
- Inadequate access controls.
- Missing security measures to stop form and user-submitted data getting compromised.
For example, a business owner might use AI to build a customer portal that lets users access invoices or account info. The portal might look professional, but if the underlying code doesn’t restrict access properly, customer info could be exposed.
What businesses should do: Get the professionals to review and test AI-generated sites and apps before you connect them to live business systems or customer data.
3. AI Tools May Have More Access Than They Actually Need
Some AI tools need permissions to access files, repositories, cloud services, or other apps. Those permissions let the tools do their job, but they also increase the risk if the tool or one of its integrations gets hacked.
Think about a business that uses an AI assistant to automate its reporting. If that assistant has access to the company’s CRM, accounting platform, and shared files, a compromised integration could put all those systems at risk.
That’s why the principle of least privilege is so crucial.
What to do: Only give AI tools the access they really need to get their job done. Review the permissions regularly and cut any connections that aren’t needed anymore.
4. Employees Might Be Using AI Without IT’s Approval
One of the biggest headaches businesses face is shadow AI, employees using AI tools without anyone knowing or giving their okay.
An employee might use an AI tool to make a summary of a customer contract, automate a spreadsheet, or help troubleshoot a technical problem. Their heart might be in the right place, but they could still accidentally share sensitive info or introduce an unapproved integration.
Without a clue about which tools are being used, businesses struggle to find the security risks or even enforce consistent policies.
What to do: Create an AI Acceptable Use Policy that spells out which tools are okay to use, what information can be shared, and when employees need permission.
5. AI Can Speed Up Development, But Not Without Some Catch
One of the advantages of AI is that it lets businesses develop applications and workflows faster than ever before. But that speed can also mean less time to do security reviews, testing, documentation, and all the ongoing maintenance that comes with it.
So, a website built with AI still needs updates, just like an AI-generated application needs access controls, and an automated workflow needs monitoring. And when something goes wrong, the business still needs to be ready to respond.
What to do: Include cybersecurity reviews in every AI implementation, not as an afterthought once the tech is already in use, but as a crucial part of the process.
How Can Businesses Adopt AI Without Increasing Their Cybersecurity Risk?
The answer isn’t to stop using AI. It offers some amazing opportunities for businesses to boost productivity, cut out repetitive work, and develop new capabilities.
The key is to adopt AI with a clear head and a good understanding of the risks. Before introducing a new AI tool, business owners need to be able to answer these questions:
- What information will the AI tool be able to access? Will it interact with customer data, financial records, employee info, or other sensitive systems?
- Who built the tool and its integrations? Evaluate the vendor, extensions, and third-party components that are part of the deal.
- What permissions does it need? Don’t give it unrestricted access when limited permissions will get the job done just as well.
- Has the tech been reviewed for security? Check out the AI-generated code, connected apps, and potential vulnerabilities.
- Who’s in charge of managing and monitoring it? Assign someone to handle updates, security alerts, access reviews, and incident response.
- Do employees know the rules? Establish clear guidelines for approved AI tools and acceptable use.
These steps can help businesses get the most out of AI while minimizing unnecessary risk.
Why Cybersecurity Should Be Part Of Your AI Readiness Plan
A lot of businesses approach AI adoption by asking “what can this tech do for me?” And that’s a fair question to start with, but it shouldn’t be the only thing on your mind.
Before implementing AI, your business also needs to figure out if your IT environment can handle it securely. An AI Readiness Assessment can help you identify gaps in areas such as:
- Existing infrastructure and technology systems
- Cybersecurity protections and potential vulnerabilities
- Data access and information security
- Application integrations and permissions
- Employee AI usage and internal policies
- Opportunities to implement AI responsibly
Taking the time to evaluate these areas before deployment can help you avoid costly mistakes and make better tech decisions.
How Enitech Helps Businesses Implement AI Securely
At Enitech, we believe businesses should be able to take advantage of new tech without introducing unnecessary risk. Our team helps orgs evaluate their IT environments, identify security gaps, and develop practical strategies for adopting new tech.
Through our AI Readiness services, we help businesses understand where AI can deliver value and what safeguards they should consider before implementation.
We also offer managed IT services, cybersecurity solutions, security assessments, and ongoing tech guidance to support businesses as their needs change.
Whether you’re building an AI-powered website, automating internal processes, or integrating AI into everyday ops, having the right tech strategy can make a real difference.
AI Should Help Grow Your Business, Not Put It at Risk
The Claude Code extension incident is a reminder that cybersecurity risks don’t magically disappear just because a tech is new, popular, or powered by AI.
In fact, the speed at which businesses are adopting AI makes security oversight even more important than ever.
You don’t need to be a technical expert to know when to use AI. You need to get a sense of what it can see, how its being handled and whether your business has got the right defences in place.
Ready to take a closer look at how AI could benefit your business ?
Enitech’s free AI Readiness Assessment gives businesses a chance to size up their tech setup, spot any potential pitfalls and figure out the best way to move forward with responsible AI adoption.
Ready to give your business an AI check up?
Already got AI working in your business ? Well, chances are you’ve still got a bunch of questions. Why not make sure your employees know how to use it properly and keep things on the straight and narrow with our Free AI Acceptable Use Policy Generator.