The accretion disks of active galactic nuclei (AGN) are promising locations for the merger of compact objects detected by gravitational wave (GW) observatories. Embedded within a baryon-rich, high density environment, mergers within AGN are the only GW channel where an electromagnetic (EM) counterpart must occur (whether detectable or not). Considering AGN with unusual flaring activity observed by the Zwicky Transient Facility (ZTF), we describe a search for candidate EM counterparts to binary black hole (BBH) mergers detected by LIGO/Virgo in O3. After removing probable false positives, we find nine candidate counterparts to BBH mergers mergers during O3 (seven in O3a, two in O3b) with a $p$-value of 0.019. Based on ZTF sky coverage, AGN geometry, and merger geometry, we expect $approx 3(N{rm BBH}/83)(f{rm AGN}/0.5)$ potentially detectable EM counterparts from O3, where $N{rm BBH}$ is the total number of observed BBH mergers and $f{rm AGN}$ is the fraction originating in AGN. Further modeling of breakout and flaring phenomena in AGN disks is required to reduce our false positive rate. Two of the events are also associated with mergers with total masses $> 100M_odot$, which is the expected rate for O3 if hierarchical (large mass) mergers occur in the AGN channel. Candidate EM counterparts in future GW observing runs can be better constrained by coverage of the Southern sky as well as spectral monitoring of unusual AGN flaring events in LIGO/Virgo alert volumes. A future set of reliable AGN EM counterparts to BBH mergers will yield an independent means of measuring cosmic expansion ($H_0$) as a function of redshift.