You could couple this with code-coverage requirements, like coverage must always increase or stay level with every commit. That can still be gamed but at least it's harder and more obvious to other devs
This can backfire in unexpected ways. Don't forget that deduplicating tested code decreases the coverage ratio. You don't want to forbid this, though.
People would just write tests which can never fail, and (in the best case) add the assertions in the end, one by one.
Speaking of tests that never fail... I've seen contractors hand over code exclaiming they've met the 75% test coverage required by SalesForce's platform but never actually asserting anything.