An adaptive traffic signal controller (ATSC) combined with a connected vehicle (CV) concept uses real-time vehicle trajectory data to regulate green time and has the ability to reduce intersection waiting time significantly and thereby improve travel time in a signalized corridor. However, the CV-based ATSC increases the size of the surface vulnerable to potential cyber-attack, allowing an attacker to generate disastrous traffic congestion in a roadway network. An attacker can congest a route by generating fake vehicles by maintaining traffic and car-following rules at a slow rate so that the signal timing and phase change without having any abrupt changes in number of vehicles. Because of the adaptive nature of ATSC, it is a challenge to model this kind of attack and also to develop a strategy for detection. This paper introduces an innovative "slow poisoning" cyberattack for a waiting time based ATSC algorithm and a corresponding detection strategy. Thus, the objectives of this paper are to: (i) develop a "slow poisoning" attack generation strategy for an ATSC, and (ii) develop a prediction-based "slow poisoning" attack detection strategy using a recurrent neural network -- i.e., long short-term memory model. We have generated a "slow poisoning" attack modeling strategy using a microscopic traffic simulator -- Simulation of Urban Mobility (SUMO) -- and used generated data from the simulation to develop both the attack model and detection model. Our analyses revealed that the attack strategy is effective in creating a congestion in an approach and detection strategy is able to flag the attack.