Context. Thanks to the advent of large-scale optical surveys, a diverse set of flares from the nuclear regions of galaxies has recently been discovered. These include the disruption of stars by supermassive black holes at the centers of galaxies – nuclear transients known as tidal disruption events (TDEs). Active galactic nuclei (AGN) can show extreme changes in the brightness and emission line intensities, often referred to as changing-look AGN (CLAGN). Given the physical and observational similarities, the interpretation and distinction of nuclear transients as CLAGN or TDEs remains difficult. One of the obstacles of making progress in the field is the lack of well-sampled data of long-lived nuclear outbursts in AGN. Aims. Here, we study PS16dtm, a nuclear transient in a Narrow Line Seyfert 1 (NLSy1) galaxy, which has been proposed to be a TDE candidate. Our aim is to study the spectroscopic and photometric properties of PS16dtm, in order to better understand the outbursts originating in NLSy1 galaxies. Methods. Our extensive multiwavelength follow-up that spans around 2000 days includes photometry and spectroscopy in the UV/optical, as well as mid-infrared (MIR) and X-ray observations. Furthermore, we improved an existing semiempirical model in order to reproduce the spectra and study the evolution of the spectral lines. Results. The UV/optical light curve shows a double peak at ∼50 and ∼100 days after the first detection, and it declines and flattens afterward, reaching preoutburst levels after 2000 days of monitoring. The MIR light curve rises almost simultaneously with the optical, but unlike the UV/optical which is approaching the preoutburst levels in the last epochs of our observations, the MIR emission is still rising at the time of writing. The optical spectra show broad Balmer features and the strongest broad Fe II emission ever detected in a nuclear transient. This broad Fe II emission was not present in the archival preoutburst spectrum and almost completely disappeared +1868 days after the outburst. We found that the majority of the flux of the broad Balmer and Fe II lines is produced by photoionization. We detect only weak X-ray emission in the 0.5−8 keV band at the location of PS16dtm, at +848, +1130, and +1429 days past the outburst. This means that the X-ray emission continues to be lower by at least an order of magnitude, compared to archival, preoutburst measurements. Conclusions. We confirm that the observed properties of PS16dtm are difficult to reconcile with normal AGN variability. The TDE scenario continues to be a plausible explanation for the observed properties, even though PS16dtm shows differences compared to TDE in quiescent galaxies. We suggest that this event is part of a growing sample of TDEs that show broad Balmer line profiles and Fe II complexes. We argue that the extreme variability seen in the AGN host due to PS16dtm may have easily been misclassified as a CLAGN, especially if the rising part of the light curve had been missed. This implies that some changing look episodes in AGN may be triggered by TDEs. Imaging and spectroscopic data of AGN with good sampling are needed to enable testing of possible physical mechanisms behind the extreme variability in AGN.