Very interesting subject. For those readers in the United States, from the Code of Federal Regulations:
"Title 49: Transportation
PART 173-- SHIPPERS--GENERAL REQUIREMENTS FOR SHIPMENTS AND PACKAGINGS
Subpart G--Gases; Preparation and Packaging
§ 173.301 General requirements for shipment of compressed gases and other hazardous materials in cylinders, UN pressure receptacles and spherical pressure vessels.
(a) General qualifications for use of cylinders.
...
(8) The pressure of the hazardous material at 55 °C (131 °F) may not exceed 5/4 of the service pressure of the cylinder. Sufficient outage must be provided so the cylinder will not be liquid full at 55 °C (131 °F)."
Electronic Code of Federal Regulations:
Note that in this usage, regular old air becomes a "hazardous material" when compressed. It's all about potential energy... I didn't see anything in that part that mentioned any other pressure restrictions. I also could not find a definition of "liquid full."
It seems to me, the bottom line is that if you're filling an AL80 to 3000 PSI, you're cramming 77.4 cubic feet of air (at ~14.7 PSI) into a space less than half of a cubic foot (at ~3000 PSI). That's inherently dangerous. If you cram in more than that much air ("overfilling"), it's more dangerous; if you cram in less air ("under filling"), it's less dangerous. If you keep cramming air into the tank, eventually it WILL fail, likely catastrophically. And you don't know at exactly what pressure it will fail. Is it possible to fill a tank to significantly more than its service pressure without failure? Sure. We've all gotten AL80s that read more than 3000 PSI at 70F. Is it possible that a tank will fail at less than it's service pressure? Absolutely.
You can dissect the numbers all you want, but if YOU'RE the one filling the tank, it comes down to how much risk YOU are willing to accept.
Tim