Autonomous flight termination systems are not "AI". It uses an on-board GPS and INS to figure out where the rocket is. It applies a pre-defined set of rules to the state vector and if any one of the rules fail it terminates the flight. You can read more about them here: https://www.gps.gov/cgsic/meetings/2019/valencia.pdf
*old man shouts at The Cloud*
An other question if you don’t mind: Did you ever used software which was not AI in your view?
Worth mentioning that the previous state of the art solution relied on a radio link too. Not sure if it was an implementation where jamming could led to flight termination, or where jamming could lead to failure to terminate a flight. But jamming, and resistance to it, was a concern even before autonomous flight termination.
https://en.wikipedia.org/wiki/Crew_Dragon_In-Flight_Abort_Te...
No. The shuttle broke up when the overall stack became unstable due to the right hand SRB separating because a strut that attached it to the external tank failed (due to a blowtorch effect from a failed O-ring). The Challenger orbiter ended up 'on top' and broke into several chunks because of the aerodynamic stress (one of these chunks was the crew compartment). The SRBs were destroyed by their FTS systems, but this was more than 30 seconds after Challenger broke up. The ET simply disintegrated.
The point is that rockets carrying crew do indeed have FTS systems - presence of a crew doesn't negate that need.
That doesn't sound very automatic.
I expect these things are only on test flights indeed.