AI Applications Are Your Coworker Now, But Can You Trust Them?

September 16, 2024

8 Mins Read

Table of Contents

Remember the last time you used ChatGPT to draft an email or write a report? It’s a common scenario now, isn’t it? AI has seeped into our workplaces, transforming the way we work. But as it becomes more ingrained in our professional lives, a crucial question arises: are we ready to trust AI applications to lead the way?

AI applications are rapidly becoming indispensable in modern workplaces. They offer the promise of increased efficiency, improved decision-making, and reduced costs. However, this technological revolution is not without its challenges and risks that one must carefully navigate.

The Rise of AI Applications in the Workplace

AI applications are being deployed across a wide range of functions in the workplace, from virtual assistants like generative AI tools to advanced software like Copilot for Microsoft 365, ChatGPT, and Gemini. According to ABC News, a majority of Australian workplaces are already deploying AI applications and tools.

Moreover, according to the 2024 Work Trend Index Annual Report, the use of generative AI has experienced a substantial increase in the past six months, with 75% of workers actively employing it and 78% proactively bringing their own AI tools into the workplace.

According to Microsoft, employees report that AI has enabled them to save time (90%), focus on critical tasks (85%), enhance their creativity (84%), and derive greater job satisfaction (83%).

On the other hand, Gallup’s Workforce Study revealed that most U.S. employees — seven in 10 — never used AI in their job, with only one in 10 saying they use AI on a weekly basis or more often. Among the 10% of employees who frequently use AI, four in 10 use it to carry out routine tasks, three in 10 to learn new things, and one-quarter to identify problems.

According to a recent survey from HR software company Workday, 62% of business leaders are welcoming the use of AI applications in the workplace. There’s no denying that AI applications are designed to streamline tasks, enhance communication, and improve efficiency.

Despite its advantages, the adoption of AI applications has raised numerous questions, particularly about data security, application security, and trust. Even according to industry experts, both company leaders and workers have anxiety and trust issues when it comes to AI applications. Not only that, Kolide’s survey revealed that 89% use some form of AI to do their jobs at least once a month, but among them, very few are actually aware of the security risks.

Can We Trust AI With Our Sensitive Information?

Privacy concerns are one of the most significant issues when it comes to the use of AI in the workplace. As Raheel Najmi, CEO of Branex, told PureVPN, “Privacy risks in AI is another important issue that needs to be dealt with as AI systems handle vast amounts of personal sensitive data. The most simple and effective strategy to mitigate exposure to your private information is to limit data collection to what is strictly necessary for the AI to function properly.”

According to a global study, only one in two employees are willing to trust AI applications in the workplace, with their trust levels varying based on their job role, country of residence, and the specific use of AI.

So the pressing issue is AI privacy, where numerous employees and businesses remain skeptical about how AI companies handle their data. Some fear that data collected by AI applications could be used for purposes beyond their original intent, such as training other models or being shared with third parties without consent.

Along with this, Microsoft research reveals that 79% of managers agree that their company needs to adopt AI to stay competitive, but 59% worry about quantifying the productivity gains of AI. In the end, AI companies must prioritize application security and data security to prevent security breaches. However, even the most advanced systems are not immune to exploitation, and as AI applications continue to evolve, so do the threats they face.

Pranjali Ajay Parse, a data scientist at Autodesk, shared similar insights with PureVPN regarding the complexities of maintaining data privacy while using AI.

“Through my work building digital tools aimed at reducing burnout and stress, I have navigated the complexities of data pseudonymization and adhered to strict privacy regulations,” she said.

Unauthorized Use of AI Applications

Employees at many organizations are engaging in widespread use of unauthorized AI applications.

Andy Bissonette, marketing director at Liquid Web, shared an example with PureVPN, highlighting the real dangers of mishandling AI systems.

“A colleague once used an AI tool to draft a document that contained client-specific information. Months later, the AI inadvertently referenced details from that interaction in a completely different context,”he said. “It was a wake-up call that reinforced the need for more rigorous oversight when using these tools.”

This underscores how unauthorized use of AI applications can store and reuse data, sometimes in unintended ways, presenting both operational and security risks.

About 74% of the ChatGPT use at work is through non-corporate accounts, potentially giving the AI the ability to use or train on sensitive data like customer details, employee records, and even proprietary corporate data.

All this leads to increased risk of data breaches, and if AI applications aren’t properly secured, businesses could face devastating consequences. Hackers can also exploit weaknesses in AI systems to manipulate them or steal confidential data.

Vulnerabilities of AI Applications in Workplace

Businesses are being left vulnerable to a range of cybersecurity and privacy risks as 70% of business executives prioritize innovation over security when it comes to generative AI projects, according to a new report by IBM.

AI applications are vulnerable to manipulation through data poisoning, where hackers feed false data into the system, compromising its accuracy and effectiveness. The reliability of AI in these high-stakes scenarios remains a contentious issue.

Moreover, in May, privacy advocates labeled Microsoft’s new Recall tool as a potential “privacy nightmare” because it can capture screenshots of your laptop every few seconds. The tool has drawn the attention of the UK’s Information Commissioner’s Office, which is now urging Microsoft to provide more details about the safety of the product set to launch in its upcoming Copilot+ PCs.

Besides this, there are growing concerns surrounding OpenAI’s ChatGPT, as its upcoming macOS app has shown the ability to also take screenshots—raising alarms among privacy experts who warn this could lead to the capture of sensitive information and could pose potential privacy infringement risks.

The U.S. House of Representatives has banned its staff from using Microsoft’s Copilot after the Office of Cybersecurity identified it as a security risk, citing concerns about leaking House data to non-approved cloud services.

Additionally, market analyst Gartner has warned that using Copilot for Microsoft 365 poses risks of sensitive data exposure both internally and externally. Last month, Google also had to revise its new AI Overviews search feature after viral screenshots showed strange and misleading responses to queries.

The Risks of Sensitive Data Exposure

Using generative AI in the workplace brings significant risks, especially around exposing sensitive data. Generative AI systems absorb vast amounts of information from the internet to train their models, which raises concerns about data privacy.

As organizations adopt AI systems, the potential for security breaches grows, particularly when sensitive data is involved.

“While generative AI and AI assistant platforms should be safe and secure, you can’t be certain of that fact,” Edward Tian, CEO of GPTZero, told PureVPN. “If you share personal or sensitive information with these kinds of platforms, you typically cannot know if or where that information is being shared. Additionally, these platforms, when connected to your other work platforms, present a new potential location for cyberattacks to occur.”

Most generative AI systems are essentially big sponges, soaking up huge amounts of information from the internet to train their language models. So these generative AI tools, due to their vast data absorption, risk storing and potentially leaking sensitive information could result in data breaches.

AI models, such as large language models (LLMs), could be targeted by hackers to access private data or compromise the system to spread false outputs or malware. Besides this, AI tools like Microsoft Copilot could be exploited by employees to access restricted information, increasing the risk of internal data leaks.

“Working with AI in email marketing is like having a
super-smart intern who sometimes needs a reality check.
There was this one time when we were playing around with an AI
for content ideas. It started spitting out stuff that was so spot-on,
it was almost creepy. Got us thinking hard about where it was pulling
this info from and how much it was ‘understanding’ about our
customers,” Scott Cohen, CEO of InboxArmy, told PureVPN.

His humorous yet insightful strategy of using a “digital disguise” with AI reflects how cautious one must be when dealing with potentially sensitive customer data.

Best Practices for Using AI Tools Safely and Effectively

Generative AI tools can enhance productivity but also carry privacy and security risks if used improperly.

Chris Lyle, co-founder of CompFox, an AI-enhanced legal research platform, told PureVPN, “Personally, I’m careful sharing details on public platforms. I aim to provide value to readers without compromising client privacy. For example, when discussing case studies, I avoid specifics that could identify clients if accessed improperly.”

His approach is a model for others looking to balance productivity gains from AI with stringent privacy measures. Here are steps businesses and individual employees can take to improve privacy and security when it comes to using AI applications in the workplaces:

Avoid Sensitive Data Input: Refrain from inputting confidential information into generative AI prompts.
Craft Generic Prompts: Use generic language in AI prompts to maintain privacy, adding specific details later.
Validate AI Output: Cross-check AI-generated content, particularly research and code, before using it.
Configure AI Systems Properly: Follow the principle of least privilege and ensure proper setup to prevent unauthorized data access.
Manage Data Usage Settings: Be aware of how AI tools like ChatGPT handle your data and adjust settings accordingly.

Security isn’t just about limiting data input; it’s also about encryption and controlling access.

“One of the best ways to reduce data breaches is through strong
encryption practices, ensuring that even if data is intercepted,
it remains unreadable, Deepak Shukla, CEO of Pearl Lemon Group,
told PureVPN. “Additionally, segmenting sensitive data and limiting
access ensures that employees and AI systems only interact with what’s necessary.”

His advice highlights the importance of combining AI innovation with robust security measures.

Should You Trust AI Applications?

So, should you trust the growing presence of AI applications in your workplace? The answer depends on how well the risks are mitigated. AI applications offer incredible potential to revolutionize the workplace by increasing efficiency, automating mundane tasks, and providing insights based on data analysis.

But without strong security measures, privacy protocols, and human oversight, these benefits could be overshadowed by significant risks and their success will ultimately depend on how well organizations address the cybersecurity challenges.

On a side note, for more exciting product launches, insightful security updates, and exclusive deals, be sure to follow PureVPN Blog.

Have Your Say!!