The first law is specifically framed in terms of 'closed systems', yet what constitutes full thermodynamic enclosure is always open to question. Fundamentally, the system has to be open to a fundamental force constant, and time. That could be the EM force constant, alpha, or the gravitational constant G etc., these reducing to effective time rates of exchange of momenta or ±dp/dt, in turn opening the possibility of divergent inertial reference frames and hence a breakdown of conservation of energy between velocity frames. OU systems more generally are 'inadvertently / inexplicably open thermodynamic systems', where one might superficially expect them to be closed and isolated, such that ie. calorimetry would be defeated. Historically, Bessler's wheel was a legit claim, repeatedly demonstrated and accredited to the highest standards; i suspect Rossi's eCat is also legit, probably Ylidiz too, but as ever the challenge is figuring out HOW and why the effect is being generated, taming what most dismiss as an invisible pink elephant..
The whole field's still marred in conflicted thinking, on both sides.. 'perpetual motion' simply Newton's first law, pending some external force acting to change it.. OU - together with under-unity (AKA non-dissipative loss mechanisms) - are a spectrum-condition of novel I/O force / space / time asymmetries, as i say, pivoting on fundamental force constants and time.. That is, to play the energy game you first have to play the momentum game, challenging N2 (F=mA and its inversions) and N3 (instantaneous equality of momenta and counter-momenta). OU / UU means a divergent inertial frame; that is, one proceeding without inertial interaction with its environment, and this applies equally to classical EM theory as mechanics. In other words caveat emptor - there's always a corresponding entropy change, somewhere.. what matters is that the worst effects of any resulting fallout are anticipated and mitigated. No free-energy panacea, no actions without consequences. The vacuum is obviously not 'nothing', and engineering it is not something you wanna be doing inadvertently (see how big-rip scenarios are contingent on localised variations in the strength of the Higgs interaction triggering runaway equilibration of false-vacuum potentials and collapse into a lower or true-vacuum state, for instance). Bessler's five-week demonstration of his largest, most-powerful wheel at castle Weissenstein through winter 1717 is coincident with the Christmas storms of 1717 that devastated the NW European coastline weeks later; resting momentum states are not disturbed lightly or trivially..