From Digital Footprints to AI Footprints
In our previous blog, we explored digital footprints, the trail of information we leave behind through our online activity. From social media posts and public profiles to websites, usernames, searches, and other online interactions, these seemingly small pieces of information can come together to reveal a surprisingly detailed picture of a person.
But our digital footprint is no longer limited to what we post publicly or what others can find about us online. AI is becoming part of that footprint too.
Artificial intelligence has quickly moved from something we experiment with to something we rely on every day. People use tools such as ChatGPT, Claude, Gemini, and Microsoft Copilot to write emails, solve work problems, analyse documents, plan trips, research topics, develop business strategies, answer health questions, and even navigate personal relationships and everyday decisions.
And unlike many parts of our traditional digital footprint, our AI conversations can contain information we have voluntarily shared in private. That raises an important question:
What happens when all of those conversations begin to tell a story about you?
A single prompt might reveal very little. But hundreds of conversations accumulated over months or years could potentially reveal names, employers, colleagues, locations, travel plans, family relationships, interests, projects, concerns, routines, health problems and future plans.
In other words, your AI account may be becoming something much more significant than another app on your phone.
It is becoming an extension of your digital footprint.
Your AI Knows More Than What You Type
When we think about AI privacy, it is easy to focus only on prompts. But increasingly, people are giving AI access to much more.
Users upload PDFs, spreadsheets, screenshots, photographs, contracts, CVs, reports, presentations, meeting notes, and other documents. Some AI services can also connect with other applications and sources of information.
For example, Google’s Gemini documentation describes connected-app functionality involving information such as emails, files, events, photos, and videos. This means the potential picture isn’t limited to what you consciously type into a chatbot.
Consider what might exist inside an AI account after years of regular use:
- Work emails and meeting notes
- Business plans and project information
- Travel itineraries
- CVs and professional history
- Personal photographs
- Financial questions
- Family or relationship discussions
- Research and interests
- Customer or client information
- Draft communications
- Future plans and decisions
Individually, these pieces may not seem particularly revealing. Together, they can provide context. And context can be extremely valuable.
AI Conversations Can Reveal the “Why,” Not Just the “What”
An email address is useful information. A telephone number is useful information. Knowing someone’s employer or job title can also be useful.
But imagine knowing why that person is travelling, who they work with, what project they are currently handling, what problem they are worried about, and what they are planning to do next.
That is a very different level of intelligence. This is where AI accounts become particularly interesting from an OSINT perspective.
Traditional Open-Source Intelligence involves collecting fragments from different sources: social media profiles, websites, usernames, public records, online publications, breached information, and other publicly available data, and connecting those fragments to build a picture of an individual or organisation.
AI potentially reverses that process.
The user may have already assembled the fragments themselves. And they may have assembled them in one place.
Over time, an AI conversation history could potentially provide enough context to understand someone’s professional life, interests, contacts, routines, priorities, and decision-making.
What If Someone Accessed Your AI Account?
Here is a simple thought experiment:
If someone logged into your AI account today, what could they learn about you?
Depending on the account, conversation history, uploaded files, connected applications, and sharing settings, the answer could be surprisingly detailed.
An attacker doesn’t necessarily need access to your entire digital life to make use of the information.
A handful of conversations might reveal:
- Your employer and role
- The names of colleagues or clients
- Upcoming travel
- Current projects
- Internal terminology
- Personal interests
- Important dates
- Family relationships
- Problems you’re trying to solve
- Organisations you interact with
- What you are planning next
That information could potentially make social engineering, impersonation, phishing, or targeted attacks more convincing.
A generic phishing email is easy to ignore, but a message that references a real project, colleague, upcoming trip, or business problem can feel much more believable.
This is why protecting an AI account should be viewed as part of personal and organisational security, not simply privacy housekeeping.
When Personal and Corporate Information Collide
There is another risk that organisations need to consider.
Employees increasingly use AI to make their work faster. They may paste meeting notes into a chatbot, ask it to summarise an internal email, upload a document for analysis, or provide code for troubleshooting.
The intention may be completely innocent. But the information could still be sensitive.
Customer information, intellectual property, internal strategies, source code, financial information, contracts, and confidential communications can become exposed if employees put them into AI services without understanding the organisation’s policies or the platform’s data controls.
The issue is not that AI is inherently unsafe. The issue is understanding what happens to information after you provide it.
Different AI platforms, account types, and configurations can have different retention, sharing, training, and privacy controls. Enterprise environments may also provide different protections from consumer accounts.
That is why organisations should establish clear policies around what employees can and cannot enter into AI systems.
Sharing Features Can Create Another Blind Spot
AI conversations can also be deliberately shared.
A user might create a link to a conversation to send it to a colleague, friend, or client and later forget that the link exists.
ChatGPT provides controls for reviewing and managing shared conversations through its Data Controls. Google similarly describes Gemini’s chat-sharing functionality as creating a public link. The important lesson is simple:
If you have shared an AI conversation, treat that link as another potential exposure point.
Old shared conversations are worth reviewing periodically, particularly if they contain personal, professional, or sensitive information.
So, Is It Safe to Share Information With AI?
There is no universal yes-or-no answer.
AI services differ in how they handle data, including retention, model training, account controls, sharing, connected applications, and enterprise protections.
Some services provide settings that allow users to control whether conversations can be used to improve models. Others offer business or enterprise environments with additional administrative controls.
The important thing is not to assume that every AI service works the same way. Instead, ask:
What does this particular service do with my information?
Before using an AI tool, understand its privacy settings, data controls, retention policies, and sharing options. If you’re using AI for work, also understand your organisation’s policies and whether the account you are using is approved for sensitive information.
Privacy should be based on understanding, not assumptions.
Practical AI OPSEC
You don’t need to stop using AI to protect yourself. Instead, treat your AI account like an important part of your digital identity.
A few practical steps can make a significant difference:
- Use a strong, unique password or passkey. Don’t reuse credentials from other services.
- Enable multi-factor authentication (MFA) where available.
- Review active sessions and connected devices and remove anything you don’t recognise.
- Review shared conversation links and revoke access where appropriate.
- Understand your AI service’s data and training controls.
- Delete conversations you no longer need, particularly those containing sensitive information.
- Think carefully before uploading documents. A screenshot or PDF can contain far more information than you realise.
- Avoid entering confidential corporate information into personal AI accounts.
- Follow your organisation’s AI and data-handling policies.
- Review connected applications and permissions and remove access that you no longer need.
Most importantly, change the mindset. Your AI account is not “just another app”, it may contain a detailed record of the things you work on, worry about, plan, research, and care about. It is something worth protecting.
Your AI History is Part of Your Digital Footprint
AI is incredibly useful, and responsible use can bring enormous benefits to individuals and organisations. The goal is to understand that the way we interact with AI is creating a new form of digital footprint.
We spend a lot of time thinking about what the internet knows about us. Perhaps we should also start thinking about what our AI knows about us and who could access it.
At The OSINT Group, we help organisations understand digital footprints, information exposure, and the intelligence that can be derived from seemingly disconnected data. If you’d like to understand what your organisation’s digital footprint may be revealing, get in touch with our team to discuss how we can help.
