The availability of ICT services can be severely disrupted in the aftermath of disasters. Ad-hoc assemblages of communication technology have the potential to bridge such breakdowns. This article investigates the use of an ad-hoc system for sending SOS signals in a large-scale exercise that simulated a terrorist attack. In this context, we found that the sensitivity that was introduced by the adversarial nature of the situation posed unexpected challenges for our approach, as giving away one's location in the immediate danger of a terrorist attack became an issue both for first responders and the affected people in the area. We show how practices of calling for help and reacting to help calls can be affected by such a system and affect the management of the visibility and validity of SOS calls, implying a need for further negotiation in situations where communication is sensitive and technically restrained.