DLP for ChatGPT &
Generative AI Tools
1 in 4 employees paste data into AI tools they should not. Aona stops it, in real time. Employees use ChatGPT daily for productivity. Without DLP controls, confidential data flows freely into AI systems outside your control. Aona detects, redacts, and blocks it, across every AI tool your team uses.
of employees have pasted confidential customer data into a public AI chatbot, where it can be retained by the provider and used to train models.
What Gets Leaked into AI Tools
These four data types appear most frequently in AI prompts, and create the most significant compliance and competitive risk.
Customer PII
Names, email addresses, phone numbers, account details, pasted into AI tools during customer service, sales, or data analysis tasks.
Financial Records
Revenue data, forecasts, earnings reports, payment card numbers, and budget documents shared with AI tools before public disclosure.
Source Code
Proprietary algorithms, API keys, database credentials, and internal codebases submitted to AI coding assistants without restriction.
Internal Strategy Docs
Board minutes, M&A information, competitive strategy, and legal correspondence used as AI context for drafting or summarisation.
How Aona DLP Works
Three-stage pipeline from prompt to protection, all in real time, before data leaves your environment.
Real-Time Prompt Inspection
Aona intercepts every prompt before it is submitted to an AI tool. The content is inspected in milliseconds, no latency impact on the employee experience.
Classification
Aona classifies the prompt content against your data policies, identifying PII, financial data, source code, and custom data types using AI-native pattern recognition.
Block, Redact, or Alert
Depending on your policy, Aona blocks the prompt, automatically redacts the sensitive portion, or alerts the employee and your security team, all in real time.
Supported AI Tools
Aona DLP works across every major AI tool your employees use, and the ones they're using without telling you.
ChatGPT
OpenAI
Microsoft Copilot
Microsoft
Google Gemini
Claude
Anthropic
Perplexity
Search AI
Mistral
Mistral AI
DeepSeek
DeepSeek AI
+ 5,000 more
All AI tools
How Aona DLP for ChatGPT Works
Real-time prompt inspection, protecting data before it ever leaves your organisation.
Deploy in Minutes
The Aona browser extension rolls out via MDM in under 5 minutes. No endpoint agents, no VPN, no firewall changes. Works across ChatGPT, Copilot, Gemini, Claude, and 5,000+ other AI tools from day one.
Inspect Every Prompt
Aona intercepts prompts before they're sent to AI models. Every prompt is scanned for PII, financial data, source code, healthcare records, and custom data types, in real time, with zero latency impact.
Block, Redact, or Alert
Depending on your policy, Aona blocks sensitive prompts, automatically redacts the sensitive portion and lets the rest through, or alerts the employee and your security team. Full audit log captured for every event.
of employees use AI tools not sanctioned by IT or security
Microsoft WorkLab, 2025
average cost of a data breach
IBM Cost of a Data Breach Report, 2024
Australian Healthcare College
An Australian healthcare college had approved Microsoft Copilot, but staff kept pasting content into consumer AI tools with no visibility or audit trail. Aona's real-time guardrails steered staff back to the approved tool and cut Shadow AI prompts by 92.9% in three months, while keeping approved AI available for the people who needed it.
“Aona gave us visibility into which AI platforms were being accessed across the college and helped us proactively discourage use of unapproved tools while reinforcing Copilot as our approved option. It has been easy to deploy, lightweight for end users, and a valuable addition to our AI policy.”Senior Systems and Security Administrator, Australian healthcare college
ChatGPT Enterprise, Team, and free behave very differently with your data
Which ChatGPT plan an employee is signed into changes how their data is handled. Most organisations have all three in use at once, which is exactly where the exposure starts.
Free and Plus
Consumer accounts. By default, conversations can be used to train and improve models unless the user manually turns that off in settings. Nothing stops an employee signing in with a personal account on a work device and pasting client data into it.
ChatGPT Team
Business plan where, per OpenAI, conversations and content are not used to train models by default. Useful, but it only covers the workspace the company controls. It does nothing about employees using free or personal ChatGPT alongside it.
ChatGPT Enterprise
Adds SSO, longer context, and admin controls, with content excluded from training by default. Strong on governance inside the tenant, but it cannot see or stop what staff paste into ChatGPT accounts outside that tenant.
Aona enforces one consistent policy regardless of plan. Whether an employee is on Enterprise, Team, a personal Plus account, or free ChatGPT, the prompt is inspected on submit and sensitive data is blocked or redacted before it ever reaches OpenAI.
Custom GPTs and memory are the two settings most teams overlook
The risk in ChatGPT is not only the main chat box. Custom GPTs and the memory feature both create their own path for data to leave your organisation.
Custom GPTs
Employees build and share custom GPTs to automate repetitive work, and they routinely paste real customer records, spreadsheets, or internal documents in as instructions or knowledge files. That data sits inside the GPT and can be exposed to anyone the GPT is shared with. Aona inspects the content going into a custom GPT the same way it inspects a normal prompt, so sensitive data is caught before it is baked in.
Memory and training settings
ChatGPT memory can persist details across conversations, and on consumer plans chat history may feed model training unless an employee has switched it off. You cannot rely on every employee configuring those settings correctly. Aona removes that dependency by stopping sensitive data at the point of submission, so it is never stored, remembered, or used for training in the first place.
DLP for other AI assistants
Aona applies the same real-time, server-side DLP across every major assistant your team uses. Explore the tool-specific guides.
DLP for Microsoft Copilot
Cover Microsoft 365 Copilot, Copilot in the browser, and the Copilot desktop app intercepted by Aona's native endpoint.
DLP for Google Gemini
Cover Gemini on the web, Gemini across Google Workspace, and Google AI search with one consistent policy.
DLP for Claude
Cover Claude on the web and the Claude desktop app intercepted natively, including Projects and MCP connectors.
Turn DLP policy into a real AI governance program
Use these templates and resources to define what data can reach AI tools, document controls, and create the audit trail security reviewers expect.
AI Security Audit Checklist
Audit AI tool usage, data flows, prompt controls, and governance gaps before sensitive data leaves your organisation.
AI Incident Response Playbook
Give security teams a repeatable process for investigating AI data leakage, prompt abuse, and policy violations.
Find Your AI Governance Fit
Answer 5 quick questions to see which AI security and governance tools best match your stack, team size, and compliance requirements.
Frequently Asked Questions
Stop AI Data Leaks Today
Start your free 90-day trial. No VPN. No agents. Full DLP coverage for ChatGPT, Copilot, Gemini, and 5,000+ AI tools, in under 5 minutes.