The invention discloses a collaborative data anti-leakage method and system. The method comprises the following steps: collecting network traffic data generated by a host, and establishing and training a sensitive data model by combining a data anti-leakage strategy; capturing, by a terminal, the network traffic data generated by the host, and redirecting and sending the network traffic data to anoptimal access node; determining, by the access node, whether to establish an encrypted communication link with a target address according to the network traffic data; selecting, by the access node,an appropriate detection node according to the current operation status of a detection node and other reference data, and issuing a detection task; identifying, by the detection node, sensitive data in real time by combining a detection strategy with the sensitive data model, and feeding back the sensitive data to the access node; deciding, by the access node, whether to manage and control the network traffic data by combining the feedback result and a management and control strategy. By adoption of the technical scheme of the invention, the external sensitive data can be identified, managed and controlled in time, and the security protection ability of the data anti-leakage system on the data is improved.