Is my business data safe when using AI?
Updated: Oct 7
When you typed your first question into ChatGPT or Claude, nobody handed you a warning label. But here's what was true the moment you hit send:
Everything you typed — your question, the details you included, the names, numbers, and specifics you provided to get a useful answer — went to a server owned by a technology company. It was stored. In some cases, it was used to train the AI to get smarter for future users.
That's not a scandal. It's how these tools work. But it is something every business owner needs to understand before they start pasting in customer lists, contract terms, employee information, or financial details.
Think of it this way: using a free AI tool without understanding the privacy settings is like having a conversation about your business in a crowded diner and assuming nobody is listening. The information isn't necessarily going anywhere harmful — but it's not private either.
This guide is going to fix that. Plain English, no legal jargon, and specific steps you can take today.
What Data Are These AI Tools Actually Collecting?
Before you can protect your data, you need to know what these platforms are actually doing with it. Here's the honest breakdown for the three tools most contractors are using:
ChatGPT (Made by OpenAI)
What it collects: Everything you type, your account info, your IP address, your browser and device, and your usage patterns.
The big thing to know: On the free plan and the $20/month Plus plan, your conversations are used to train future versions of ChatGPT — by default. That means what you type could influence how the AI responds to other users in the future.
What you can do about it: There's a setting buried in your account that turns this off. Go to Settings → Data Controls → toggle off "Improve the model for everyone." Once that's off, your conversations are no longer used for training.
You can also use Temporary Chat for sensitive conversations — it's a toggle at the top of a new chat. Temporary chats aren't saved to your history and aren't used for training.
The safer option for business: ChatGPT Team or Enterprise plans contractually guarantee your data is never used for training. Cost is around $25–$30 per user per month.
Claude (Made by Anthropic)
What it collects: What you type within the service. Unlike Google's Gemini, Claude doesn't pull from your broader digital life — it's limited to what you actually put into it.
The big thing to know: In September 2025, Anthropic updated their terms. When users clicked through the update, a toggle was pre-set to "On" — giving Anthropic permission to use your conversations to train future models for up to five years. If you clicked through without reading carefully, you may have agreed to that without realizing it.
What you can do about it: Go to Settings → Privacy → turn off the model improvement toggle. Check this today — don't assume it's already off.
The safer option for business: Claude Team or Enterprise plans don't use your data for training at all, by default. No settings to manage. The API version of Claude is also very privacy-protective — data is retained for only 7 days and never used for training.
Gemini (Made by Google)
What it collects: Your conversations, plus data from across your Google account — search history, YouTube activity, location data, and more. If you're using a personal Google account, Gemini has access to a much broader picture of you than the other tools.
The big thing to know: On a free personal Google account, Gemini may use your conversations to improve Google's AI. Even when you turn off activity, Google holds your conversations for at least 72 hours for safety review.
What you can do about it: Go to myactivity.google.com/product/gemini and turn off Gemini Apps Activity. You can also set auto-delete to 3 months. Note that turning off activity also turns off your conversation history.
The safer option for business: If you have a paid Google Workspace business account, Gemini treats your data the same way it treats your business Gmail — private, never used for training. This is a meaningful difference from a free personal account.
The Simple Rule That Protects Most Businesses
Before you spend an hour adjusting settings, here's the single most important rule for keeping your business data safe when using AI:
Never paste anything into a free AI tool that you wouldn't be comfortable seeing posted publicly.
That includes:
Customer names, addresses, emails, or phone numbers
Employee information, salaries, or HR details
Contract terms, pricing, or profit margins
Banking information, account numbers, or tax details
Anything a competitor would benefit from knowing
This isn't about these companies being malicious. It's about understanding that free tools are funded by data. The less sensitive information you put in, the less risk you carry — regardless of what the settings say.
A good mental test before you paste anything: "Would I be comfortable if my biggest competitor could read this?" If the answer is no, don't include it.
Understanding the Risks of Free vs. Paid Accounts
Here's where most contractors are unknowingly taking on risk — and it's worth spelling out clearly.
Free accounts have the least privacy protection. Your data is being used. Full stop.
Paid personal accounts ($20/month for ChatGPT Plus, Claude Pro, or Gemini Advanced) are NOT automatically safer. Many people assume that paying for an upgrade means their data is protected. It doesn't — not without changing your settings manually. Even if you're paying $20 per month for ChatGPT Plus, your chats are used to train models unless you manually go to Settings → Data Controls and opt out.
Business or Team accounts are where the real protection kicks in. ChatGPT Enterprise/Team, Claude Team/Enterprise, and Google Workspace Gemini all contractually guarantee that your data is not used for model training. That guarantee is in writing — not just a setting you turned on and hope stays on.
For a contracting business using AI for anything customer-facing — proposals, communications, contracts, financial analysis — a business tier account is the right move. The cost difference between a personal Plus plan and a Team plan is typically $5 to $10 per user per month. That is a very small number compared to what a data exposure could cost you.
What Happens If You Put Sensitive Data In and Then Delete the Conversation?
This is the part that surprises most people.
Even if you delete your chat history, once information has been used to train ChatGPT, Gemini, or Claude, it remains embedded in the model and cannot be fully removed.
Deleting a conversation removes it from your visible history. It does not guarantee the data is gone from the company's systems immediately — most platforms take 30 days to fully remove deleted content from their servers, and in some cases longer.
The practical takeaway: don't put something sensitive in and count on deletion to make it disappear. Prevention is the only reliable protection.
Six Practical Rules for Using AI Safely in Your Business
Here's a simple checklist. Follow these and you'll be ahead of 90% of contractors using AI today.
Rule 1: Change your privacy settings today. On every platform you use, go find the training/data setting and turn it off. It takes 5 minutes. The steps are above in this post.
Rule 2: Never paste in real customer data. If you need AI to help you write a proposal or respond to a customer, use placeholder names. Instead of "John Smith at 4532 Oak Street," write "the customer" or "Client A." You'll get the same quality output without the exposure.
Rule 3: Treat your pricing and margins as confidential. Your markup, your labor costs, your profit per job — this is the information that defines your competitive position. Don't paste it into a free AI tool. Use round numbers or percentages if you need AI to help with financial analysis.
Rule 4: Use a business-tier account for anything client-related. If you're using AI to draft contracts, respond to customer complaints, analyze bids, or handle anything that touches a real person's information — use a Team or Business plan. The contractual protection is worth the extra few dollars.
Rule 5: Be careful about what you connect. When AI tools ask permission to connect to your email, your Google Drive, your calendar, or your CRM — read what you're agreeing to. Each connection is another door. Only open the ones you actually need.
Rule 6: Create a simple rule for your team. If anyone on your team is using AI — even informally — they need to know the basics. "Don't put customer information into free AI tools" is a one-sentence policy that prevents most of the risk. Say it out loud. Put it in writing.
How to Turn Off Data Training — Step by Step
On ChatGPT:
Open ChatGPT and click your profile icon (bottom-left on web, top-right on mobile)
Click Settings
Click Data Controls
Toggle off "Improve the model for everyone"
For sensitive one-off conversations: click Temporary Chat at the top of a new chat window.
On Claude:
Open Claude and click your profile icon
Click Settings
Click Privacy
Find the model improvement/training toggle and turn it Off
On Gemini (Personal Account):
Click Turn Off at the top
Optionally set auto-delete to 3 months
Note: Turning off activity disables your conversation history.
If you use Google Workspace for business: Your Gemini data is already protected at the account level — but confirm with whoever manages your Google Workspace account that business data protections are active.
The Bottom Line - Is my data safe using AI?
Using AI tools for your contracting business is smart. The efficiency gains are real, the time savings are real, and the competitive advantage for early movers is real.
But using them carelessly with sensitive business data is a risk you don't need to take — especially when fixing it takes about 5 minutes per platform.
The businesses getting the most out of AI right now are doing two things: using the tools aggressively for the right tasks, and keeping the right guardrails in place so their customer data, their pricing, and their business strategy stay protected.
That's not a contradiction. That's just good business.
What This Has to Do With How Avidity Works With Clients
When we build automation systems and AI agents for contracting businesses through the Business Automation Blueprint™, data handling is part of the conversation from day one.
We work with business-tier accounts by default. We help clients understand what data is flowing where. And we build systems that get the efficiency gains without putting customer information or business-critical data at unnecessary risk.
If you're using AI in your business — or thinking about it — and want to make sure you're set up the right way from the start, that's exactly the kind of conversation a discovery call is built for.
Frequently Asked Questions About AI Data Safety for Contractors
Is it safe to use ChatGPT for my contracting business?
It can be — with the right settings and the right account type. On a free or personal paid plan, your conversations may be used to train future AI models unless you opt out in settings. For business use, especially anything involving customer data or sensitive business information, a Team or Business tier account provides contractual data protection. The steps to change your settings are in this post.
Does deleting a ChatGPT or Claude conversation permanently delete my data?
Not immediately. Most platforms take up to 30 days to fully remove deleted conversations from their servers. More importantly, if your conversation was already used to train the AI model, that data cannot be removed from the model even after you delete the chat. Prevention is more reliable than deletion.
What is the difference between a free AI account and a business AI account for data privacy?
On a free or personal paid account, your conversations are typically used to train future AI models by default, and you have to manually opt out. On a business or team account, your data is contractually excluded from model training — no settings required. For contractors using AI for anything involving customer or financial data, a business account is the safer choice.
Can I use ChatGPT to write customer proposals without risking my data?
Yes — with one rule. Use placeholder names instead of real customer information. Write "the homeowner" or "Client A" instead of a real name and address. You'll get the same quality output from the AI without putting actual customer data into a third-party system.
Is Claude safer than ChatGPT for business data?
Claude has historically had a more privacy-protective default position than ChatGPT. On Claude's free and Pro tiers, the company requires explicit opt-in for training rather than defaulting to it — though a 2025 terms update means you may have accidentally opted in if you clicked through without reading carefully. Check your settings. Claude Team and Enterprise accounts do not use data for training at all, by default.
What is the safest way to use Gemini for a contracting business?
The safest option is a paid Google Workspace business account with Gemini enabled. In that configuration, your data is treated like your business Gmail — private and never used for AI training. On a free personal Google account, Gemini has access to your broader Google activity and your privacy protections are significantly weaker.
What information should I never put into a free AI tool?
Customer names, addresses, phone numbers, or emails. Employee information, salaries, or HR records. Contract terms, markups, or profit margins. Banking details or tax information. Anything you would not want a competitor or the public to see. Use placeholder descriptions and round numbers when you need AI help with sensitive topics.
Does paying $20 a month for ChatGPT Plus or Claude Pro protect my data?
No — not automatically. Paid personal plans typically require you to manually change your privacy settings to stop your conversations from being used for training. The contractual data protection that comes with business and enterprise accounts does not apply to personal paid plans. Check your settings regardless of which plan you're on.
What should I tell my employees about using AI safely?
Start with one rule: don't put real customer information into free AI tools. From there, make sure anyone using AI for work has a business-tier account and understands what types of information are off-limits. A simple one-page policy is enough to cover most small contracting businesses.
How does Avidity Innovations handle data privacy when building AI systems for contractors?
We use business-tier AI accounts by default for all client work. Data handling is part of every Business Automation Blueprint™ engagement — we help clients understand what information is flowing through their AI systems and build automation in a way that protects customer and business data from the start.
Avidity Innovations is a fractional Chief Automation Officer firm serving specialty contractors, trades businesses, and B2B service companies. We help owners identify where their operation is losing time and margin — and build AI and automation systems that work safely and effectively. Headquartered in Raleigh, NC. Serving businesses nationwide. Visit us at *avidityai.com
Privacy settings and platform policies change frequently. Verify current settings directly on each platform's privacy or settings page before relying on any specific configuration.
Comments