Current Web Service security standards have inadequate support for end-to-end protection of data when some receivers of the data are unknown to the sender. This paper presents an approach to aid collaborative partner services in properly protecting each other's data. Our approach allows each partner to derive an adequate protection mechanism with minimum performance overhead for each message it sends based on those of the corresponding messages it receives. We modify the message handling mechanisms of Web Service engines to dynamically gather protection requirements for a given outgoing message by aggregating requirements from original owners of message data.