There's a case against globally accessible DNA databases from the point of view of dragnet approaches picking up people whose connection to the crime scene is either totally incidental or actually nonexistent and purely down to data mixups, but that's a largely separate issue from authoritarian police trying to massage the evidence in a case against a particular person.

If you're already a "convenient suspect" the police wish to build a case against, they don't need to find you on a consumer genetics database to be able to obtain a sample of your DNA or strongly hint that they expect to be able to use DNA evidence (truthfully or otherwise) to send you away for a very long time if you're not willing to consider moving to plea bargaining territory.

With enough corruption, they might also "find" it on the crime scene.

Exactly, while DNA may identify an individual with high precision, we have to trust that that DNA was actually associated with the location or event.

Consider a public bathroom: how much DNA is at that location? How much is relevant to the event?

Consider a corrupt lab tech: The DNA at the event might be reported as DNA from their database. Does a defense attorney get to see the untainted sample?

Consider a corrupt/hacked DNA database: The database might report that the DNA belongs to one individual when it actually doesn't. Does a defense attorney even get to see the source code?

In these scenarios, the link to any individual is either of little consequence, or fraudulent.

And in case this sounds implausible, a corrupt lab tech handled 60,000 cases in MA:

https://www.cbsnews.com/news/annie-dookhan-chemist-at-mass-c...