Working memory (WM) can be improved after repeated training sessions paired with noninvasive neurostimulation techniques. Previously, we reported that WM training paired with tDCS succeeded behaviorally by enhancing anterior-posterior theta phase coherence and reducing alpha power. Here, in two experiments we tested several theta and alpha frequencies and two transcranial alternating current stimulation (tACS) montages in an effort to shortcut WM training while preserving behavioral gains. In Experiment 1, in separate sessions participants received online tACS at two frequencies derived from the previous study with the respective goal of improving and impairing WM performance. We selected the mean group peak value theta (7 Hz) to benefit WM and alpha (11 Hz) to impair WM. Stimulation (tACS) over right frontoparietal sites (F4-P4) during 3-back WM tasks (object, spatial) produced no behavioral consequences. In Experiment 2 we stimulated at a slower theta frequency (4.5 Hz), which was also significant in our prior study, and tested whether frontoparietal or bifrontal montages would be more effective at improving WM. This experiment revealed selectively improved object WM after right frontoparietal tACS alone. In summary, one session of tACS failed to produce the magnitude or breadth of WM gains observed after 4–10 tDCS-WM training sessions. In short, despite looking for loopholes we found little tACS savings.