The question is indeterminate since the I in EMI is the product of some unintended radiation and same spectral susceptibility in some target device.
So the most generalized case is unshielded wires of high current pulses [A/us] and high voltage [V/us] slew rate and high EM fields [V/m], [A/m]
The actual spectrum of transients that are slew rate limited may not be excessive but no less problematic with IO cables acting as antenna.
The spectrum of arcs are the broadest and repeating arcs in a small gap span the frequency up to the half wavelength of the gap and then beyond with harmonics.
Cars now use carbon sparkplug cables which current limit the ignition arcs to the plugs and thus limit the spectrum that interfere with radios, in particular AM, which is a great improvement over copper wire sparkplugs of long ago.
1Hz Xenon safety flashes in a factory where high voltage tests are conducted are well known to interfere with many cell phones due to the high EMI pinging the AGC's in the receivers with broad spectral noise spanning many GHz from the ionization impulse. (Even more-so than arc-welders) This may be limited to one supplier's poor design conformance to unintended radiation.
I have conducted many product tests for radiated, conducted and susceptibility tests before release to production. ESD, impulse conducted, radiated and pulsed RF sweeps as well as radio adjacent channel and co-channel susceptibility tests in addition to routine Hipot dielectric withstanding tests. This is an essential Design phase.