Thanks for the press release link. My knowledge is outdated. So reading the linked research paper:
The 2023 experiment with the single-dish Green Bank Telescope (GBT) demonstrated that Starlink beam-forming could reduce radio interference at the primary telescope beam (the "boresight") by as much as two orders of magnitude.
This experiment required coordination between StarLink and the GBT operators at the National Radio Astronomy Observatory (NRAO).
Another NRAO telescope in North America, the Very Large Array (VLA), has been conducting experiments of its own, in coordination with StarLink and the Navajo Nation. An enclave of the Navajo Nation, the community of Alamo is northeast of the VLA site. Isolated from the main extent of the Navajo Nation, Alamo has close ties to the nearby village of Magdalena, home to many of the VLA workers. This region surrounding the VLA could benefit greatly by StarLink internet services.
The VLA StarLink tests showed that most of the interference comes from the satellite, rather than the terrestrial user terminals.
Tests to automate the process of avoiding interference are ongoing. It will involve sharing observation scheduling and telescope configuration data from NRAO to derive beam shaping modes for the StarLink satellites.
I know that the VLA RF chain has the flexibility to handle this, and that the VLA correlator was built specifically to handle interference of this nature. The design stems from the early 2000's and was informed by increasing interference from GPS satellite constellations.
The process has yet to be automated, but there's cause to believe that it can work.
(I worked on the VLA monitor and control system software design 1999-2005. Many design reviews. Shared an office with the engineer who designed the correlator configuration API.)
The government agreed to a radio quiet zone in the areas surrounding ALMA.
But now there's Starlink and other satellite constellations coming on line at an unprecedented pace.