Don’t risk your company’s data — these are the safest ways to use AI at work

AI tools have found their way into a long list of professions.

Workers tied to the technology, finance, healthcare and retail industries are among those who use chatbots such as ChatGPT, Gemini and Claude to streamline aspects of their duties or assist them with assorted assignments. Even though they’re useful for rewriting emails, analyzing data and finding alternate methods to produce the best results, there is a massive risk that comes with using AI at work: accidentally exposing your company’s most sensitive information by sharing it with AI.

Being as careful as possible when integrating AI in your workflow can save you from all the headaches that arise if you mistakenly put your company data at risk. Using a prompt that leads to a leak of internal company info, customer data or proprietary code can be catastrophic.

Article continues below

Here are the tips I recommend to help keep your workplace secure while using company-approved internal chatbots responsibly.

Refrain from copying/pasting and rely on generic descriptions

The most common knowledge to apply when using AI at work is to never copy and paste sensitive company data into a chatbot.

You don’t want to get caught in a situation where a chatbot gets caught in a data leak that also exposes your company's confidential business plans, customer/client information, financial records and internal documents to the public.

Instead of using specific company details for a prompt, it’s better to go with generic prompts to get the advice you seek. It’s not safe to paste in customer information, then ask a chatbot to analyze that customer dataset. A better way to prompt in that circumstance is to go with this option: What’s the best way to analyze customer churn in a SaaS business?

If you’re adamant about using prompts that reference your job, it’s best not to include any identifiable information in your prompt. Turning your work problem into a case study with a chatbot works, especially when you exclude names and go with terms such as “Client A” and “This company…”

Some other applicable rules you need to consider when using AI at work include:

• Disable an AI tool’s chat history and its ability to be trained based on your chats
• Turn on a chatbot’s temporary chat or incognito mode
• Never upload company documents

Prompt safely and fill-in AI-produced frameworks afterward

Safe prompting will get you the best actionable insights possible when it comes to your company’s biggest issues. The seven following prompts are the best ones worth using in this instance:

• What are common strategies to improve low employee engagement in a mid-sized company?
• Act like a management consultant. What steps would you take to diagnose declining team productivity?
• What frameworks can be used to identify bottlenecks in a business process?
• Give me a step-by-step plan to improve communication across cross-functional teams.
• What are the pros and cons of outsourcing customer support versus keeping it in-house?
• What are the common root causes of missed deadlines in project management, and how can they be addressed?
• What warning signs should a company watch for when scaling operations too quickly?

Prompting AI to build out an outline that you can later apply to your work is better than actually pasting your company data into the chatbot and asking it to build something for it.

It’s better to use an AI-generated chart to insert data into, give you optional ways to send/answer an email, draft outlines for a business proposal and brainstorm practical ideas for your work duties.

Most AI tools come with integrated plugins, such as ChatGPT’s app store full of well-known properties and Google Workspace’s usage with Gemini. It’s best to be especially careful with these connected tools and third-party apps when using AI at work — there’s an even higher chance of getting company info exposed when using a chatbot’s collection of connected tools you don’t entirely trust.

Bottom line

Using AI carelessly at work is entirely avoidable if you follow a few simple guidelines. Tools like Perplexity AI and Meta AI can be valuable additions to your workflow, but feeding them sensitive or internal company documents to guide decisions is a mistake.

Whether you’re an employee or an executive, applying safe AI practices to your daily routine is essential.

Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds. Subscribe to Tom's Guide on YouTube and follow us on TikTok.

More from Tom’s Guide

• AI scams are getting harder to spot — pay attention to these 7 warning signs
• Don't trust ChatGPT at face-value — here's the system I use to fact-check it in seconds
• I use ChatGPT every day — I stick to these 3 rules to protect my privacy