So, this sounds like a highly evolved SDR; Are there special caveats about the frequency bands it is able to transmit/receive on?
These types of antennas typically have narrow bandwidths. That's fine for some applications, but not others. In general, antennas can be scaled to a wide range of different "center frequencies", but you care about the bandwidth a lot depending on your application. (There's a relationship between your maximum data transfer rate and the frequency range over which you need to be able to operate.)
I just skimmed the paper, but it looks like they're showing modulation on the order of a half-dozen megahertz of bandwidth. That's very narrow, but it's also impressive considering the degree of [independent] control they're claiming. Note that the researchers themselves aren't heralding this as the most amazing thing ever---in fact, they don't even mention 6G in the paper (even though this type of technology is fundamentally important in both 5G and 5G). This is an interesting advance, but it's the marketing people that are making it sound ground-breaking, as is often the case.
This could be a very handy thing indeed if it is fully broad-spectrum.
You still need an RF signal to drive it (since it's just an antenna), even if it could somehow be very wideband. As it stands right now, the frequency range of that signal source is the main thing the FCC would be angry about, but that's separate from the antenna.
... Elsewhere, people commented about beamforming and security:
You don't _need_ multiple antennas to do beamforming, you just need a region of space over which you control the EM field. That's most often accomplished with multiple antennas, but you can also do it in other ways, including the sort of metasurface leaky-wave approach they're using in this paper.
As for security, if you look at the actual paper, that's more of an afterthought. Somebody on here commented that beamforming doesn't get you security, and that's partially true. However, what they're claiming here is that, due to the ability to send beams with different modulation in different directions (which is something you can't typically do easily without separate arrays), they can reduce an eavesdropping attacker's ability to listen through antenna sidelobes since the signal they'd receive there isn't the same as at the intended receiver's location. This as least means that the eavesdropper has to be located "near" the intended direction of the receiver, which often isn't the case.