77% of Employees Share Company Secrets on ChatGPT Compromising Enterprise Policies

A GBHackers article reports that 77% of employees admitted pasting company information into ChatGPT or similar platforms, highlighting a clear risk to enterprise data. The piece cites LayerX’s Enterprise AI SaaS Data Security Report 2025, derived from real enterprise telemetry across Fortune 500 companies, which indicates AI platforms have already eclipsed conventional SaaS as the primary vector for data exfiltration.

It notes that nearly half of employees surveyed have interacted with GenAI tools in daily tasks, often copying internal documents, customer lists, and financial records into prompts, while 82% of AI tool usage occurs through unmanaged accounts outside enterprise SSO and policy enforcement. The article also points out that 75% of modern organisations use a mix of sanctioned and unsanctioned chat apps, with 87% of chat traffic flowing through unmanaged accounts and more than 62% of exchanges containing PII or PCI.

It argues that traditional DLP and perimeter controls are insufficient, recommending real-time web telemetry analysis, AI-powered behavioural analytics, and zero-trust approaches to safeguard critical data. According to the piece, the insights underscore the need for enhanced visibility across managed and unmanaged accounts to detect anomalous data flows. October 8, 2025.