Project Loon balloons also show up on Flightaware, so they either have ADS-B or TIS-B.
A situation like this will almost certainly cause some congresspeople to fret and write bills that would require ADS-B on all balloons, which would be a death knell for amateur ballooning unless ADS-B (or "legacy" Mode A/C/S) transponders become significantly smaller and more affordable. Mode C/S transponders are already available in miniaturized form factors thanks to the UAS industry, and are designed to be interrogated by aircraft equipped with TCAS (i.e. all 10+ passenger aircraft) that provides pilots deconfliction commands automatically and with no ATC support. But they're still priced for industry, not amateurs.
[0] https://www.eoss.org/ Look for N991SS, N992SS, N461SG.
[1] https://www.westmarine.com/plastimo-tubular-radar-reflector-...
A radar reflector such as that, or this (https://overlookhorizon.com/product/radar-reflector/, which is ~300g) has roughly the same RCS as a small (piper cherokee) to medium (gulfstream) sized aircraft.
That being said, detection isn't everything; primary radar cannot make accurate altitude measurements, only bearing and range. While that's enough to route traffic around, it could be also mistaken for a false return.
ADS-B out is still relatively new (especially in aviation terms) so I expect we'll see this continue to evolve.
I obviously don't know which is right, but it does show that there is definitely confusion out there about the issue.
WindBorne claims "12+ days typical flight, with demonstrated capability for 75+ day missions." So 1150Wh minimum (80Ah at 4S, which is probably like 16lb.) But you're up in the atmosphere and probably need to heat that battery so... more. But we're already at 18lb additional weight... Maybe you could offset with solar panels...
But, given that the entire balloon and payload weighs 2.5lb we're already way off the edge of feasibility for an active ads-b out.
Maybe there's something that would only listen and then respond when it heard something and that would reduce the power draw. But we're needing something 2 orders of magnitude less massive.