Yes, but ours are networked through satellites or other UAVs. Line of site is the big challenge in doing deep strikes with these sorts of low cost options. I have no idea how they are doing it ..... yet.
As with my previous commentary in this thread on drone targeting and gps, I do not claim to be an expert, but if I was tasked with doing this, my options to explore might be:
1. Integration of an on board data link (LPWAN / UWB / standard WIFI / cellular / frequency hopping 2.4 or 5GHz radio) for short range comms with localised relays and then onto centralized GPS via Starlink. Depending on power consumption, you might be able to cut the relay all together and directly link to starlink.
If a relay was required, it could be either a ground based relay with a SF crew (in which case they could just fly it in directly without the Starlink enabled GPS), or a loitering 'mother' drone within say 5km (commercial, switchblade, hobbyist even). Latency would be a major issue, as would jamming, but relay of waypoint data for initial approach followed by a low grade visual feed for the last 1000m or so might be possible, allowing human guidance onto a target.
UWB might (might) offer 1cm or better accuracy short range positioning superior to traditional GPS with lower power draw, potentially allowing for a fully autonomous approach to the requisite accuracy I guess, but that's beyond my knowledge base.
The average mobile phone can do basically all that stuff, whilst the average wireless earbud can do some of it, which gives an idea on cost and weight...
2. Similar system as above, but piggybacking on local cellular service for a direct data feed to an operator. Data tracking and stream stability would be challenging, but perhaps not impossible within urban areas with 3G+ coverage. This may limit altitude and operating area significantly.
3. Pre-programmed waypoints on the onboard GPS unit as an adaptation of the autonomous functionality used on many hobbyist drones. Basically allows the onboard gyro and servos to stabilise the craft, maintain altitude, speed and bearing, then head to a given coordinate autonomously. Not sure on the accuracy, but most civilian grade units are +/-1 meter or so, for less than $100. The system on the 'Switchblade' is an advanced version of this. Unit weight is only a couple of grams.
If it were me, I'd maybe lean towards option 3 for fully independent autonomous approach with the transmit function turned off for stealth / reduced power consumption, with it turning on upon reaching a predetermined GPS coordinate to allow option 1 for final closing to target.
Either way, I'd expect that some combination of these would allow for a relatively robust system to be built for marginal cost and minimal impact to payload.