I understand that with diode-pentode and double diode-pentode valves like the EBF80, when used as an IF amplifier and demodulator, neutralization was desirable because of the feedback pathway due to capacitance between the demodulator diode anode and the pentode control grid. There was a discussion of this topic, with circuit details, in AWV Radionics #118 of 1946 March-April, page 33ff, available here:
https://frank.pocnet.net/other/AWV_Radiotronics/.
If the receiver at interest has such a neutralization circuit, then perhaps something is astray there. If it does not (e.g. omitted for cost reasons), then one suspects that oscillation-free operation might be somewhat “statistical”, dependent upon component tolerances in summation not leaning too far away from “normal”. Were this feedback path the cause of oscillation, then perhaps it could be checked by temporarily shorting the diode anode to earth, which I think would stop the oscillation (although then there would be no signal getting to the AF section).
Cheers,