Introduction
Artificial intelligence is no longer a distant idea. It is reshaping how businesses draft documents, analyse data, write code and serve customers. Yet along with the convenience of AI comes a serious threat to the confidentiality of company information that cannot be ignored if AI is to serve the public good. I have previously written on AI governance and ethics. In this article I focus on how AI endangers confidentiality and what organisations can do about it.
How AI threatens confidentiality
Confidentiality is about keeping sensitive information from being seen or used in the wrong way. AI systems, especially generative models, depend on data, and often lots of it. The more personal or proprietary the data, the greater the risk that people’s information or trade secrets will be collected, processed or shared without proper consent or protection.
The most visible incidents involve direct user input. Company employees can accidental leak sensitive data using AI applications or tools. The leaked information may include confidential source code related to new programs, notes from internal meetings, and other proprietary data.
Leakage is not limited to copy‑and‑paste. Integrations and connectors that link AI assistants to internal repositories can unintentionally expose sensitive content if access controls and data filters are weak. It is believed that AI tools can be tricked into leaking data from private channels.
Why confidentiality matters in business and at law
Confidentiality differs from general privacy because it protects competitively significant information—source code, pricing models, mergers and acquisition plans, customer database, and internal strategies—whose disclosure can cause immediate commercial harm. When such material is shared with third‑party AI vendors, companies risk losing trade‑secret status or waiving legal privileges, especially where terms of use grant broad rights to process or retain inputs.
