Optional sets passed in python persist across unittests

I believe that this is a new issue since I've been able to run this code with no issues before:

import unittest


def foo(n, seen=set()):
    if n in seen:
        return False
    seen.add(n)
    return True

class Test(unittest.TestCase):
    def test_a(self):
        self.assertFalse(foo(1))

    def test_b(self):
        self.assertFalse(foo(1))

But I'm now seeing that test_b fails since it holds onto the value of the set in the previous unittest. I was wondering how this might have started happening.