The evolution of the sixth generation (6G) of wireless will enable the full-scale deployment of systems of cyber-physical systems (CPS) such as autonomous vehicles, robots, and drones. Currently, a paradigm shift in security and trust-building approaches for 6G is being discussed within the communications and networking communities, to tackle novel and more sophisticated behavioral attacks of malicious CPS agents. As an example, such autonomous agents will use vast amounts of data – generated from and exchanged amongst heterogeneous network devices – for artificial intelligence-driven decision-making. As a result, standard device authentication approaches and current agent reputation models will not suffice for trust building; the verification, traceability, and trustworthiness of the exchanged data themselves become critical issues. In this paper, we highlight related open issues using as an example a specific behavioral attack in a vehicular ad-hoc network (VANET), to open up the broader discussion regarding potential trust issues in 6G CPS networks. In light of this, we propose a comprehensive framework for trust and reputation management in 6G. We discuss alternative framework architectures and provide high-level details of the framework components with involved technologies.