Over the past decade it has become clear that transposable elements not only make up the bulk of eukaryotic genomes and that their continuous proliferation has had a profound impact on the biology and evolution of species. Notably, examples now abound of TE sequences coopted to assemble new genes or regulate gene expression, thereby fueling evolutionary novelty. However, in nearly all cases of TE cooption, only parts of individual TE sequences have been repurposed for cellular function and the coopted TE is no longer capable of transposition. Thus, the dogma remains that transpositionally active TEs are best viewed as selfish genetic elements not only providing no direct benefit to the host, but actually threatening genome integrity. In this talk, I will challenge this view and present evidence that active TEs can engage in activities indispensable for organismal development. We focus on two families of endogenous retroviruses in zebrafish called Bhikhari (Bik-1 and Bik-2) with nearly identical and insertionally polymorphic copies suggesting that both have been transpositionally active in the germline very recently. Interestingly, both families have evolved distinct and precise expression patterns during embryonic development: Bik-1 is expressed in the mesoderm lineage while Bik-2 (also known as crestin) marks the neural crest lineage. Manipulative experiments in zebrafish embryos show that Bik-1 and Bik-2 express Gag proteins essential for proper mesoderm and neural crest development, respectively. Mechanistically, Bik Gag proteins likely promote cell adhesion and/or oppose cell migration in a cell-autonomous fashion. Moreover, a distinct endogenous retrovirus in chicken (ERNI) encodes a Gag protein with similar activities in neural crest cell development. Thus, on at least three separate occasions, the Gag proteins encoded by active TEs have become required for vertebrate development. We propose that such interactions arise progressively during evolution as a form of addiction to TE products, in this case Gag proteins. We hypothesize that these TE products introduce cellular activities that are redundant with and occasionally replace host-encoded products essential for embryonic development, such as those controlling cell adhesion/migration. This creates dependency on the TE product for host fitness, which in turn promotes the maintenance but limit the propagation of the TE in the germline. Our data support the provocative idea that TE activities are inextricably intertwined with organismal development, as envisioned by Barbara McClintock but largely dismissed for more than half a century.