But is it safe to have 1.4 vcore?
Depends on what you are running. The more current you draw, the more vcore you can give it without promoting degradation. The only solid information we have right now comes from The_Stilt.
If you are running something AVX2-intensive (like Prime95, Linpack, or anything that behaves similarly), then the max "safe" vcore is around 1.325v with variations based on the CPU.
For some non-SIMD applications that are lightly-threaded or single-threaded, vcore up to 1.475 may be safe.
So you really need to look at current draw and compare it to something like Prime95 @ 1.325v. Bear in mind that current draw does go up with increased voltage, so be careful.
Also, I respect the car radiator OC! Too bad "big water" doesn't seem to help Matisse much. I'm thinking chilled mineral oil might be a better way to go unless you're willing to go more extreme.
edit: here is some sample data from my 3900x, which is basically 2xR5-3600.
Prime95 is the most intense workload I have run thus far. Small FFTs. Running this benchmark at 1.2875v (medium LLC to prevent vdroop), I get peak current of 135a according to HWiNFO64 and 138a according to Ryzen Master. Sustained current seems to be around 130a.
So that should give you some idea of how much current you can push @ 1.325v . Maybe a little bit more since raising vcore above my limit above would raise draw as well. I can't test Prime95 SmallFFTs @ 1.325v since my system would probably hit the 95C temp limit. I get 77C in the above test. That's with a custom water loop using a MO-RA3 (slightly smaller than your rad, I suspect). I am guessing the theoretical limit of 1.325v is achieved somewhere around 140-145a. You will not reach that without some kind of active cooling.
CBR20 which uses Embree with "some AVX code" @ 1.325v (Medium LLC) showed peak current of 106a. I'm guessing some more vcore would be "safe" here but it's impossible for me to say how much more . . . current draw tracks with voltage. If I raise vcore to 1.35v (for example) then current peak goes to 112a in this benchmark. Temps are still lower than with Prime95 SmalFFTs though.
CBR15 uses . . . SSE4.1a? No AVX. Running 1.325v (Medium lLC) showed peak current of 103a.
That's should give you some ideas. Again, be careful! And if you see something pushing your CPU to 95C or near it when above 1.325v then maybe you should back off! Nothing should be making you run hotter than Prime95.