The author reveals vulnerabilities in existing Model Ownership Resolution schemes, highlighting the risk of false claims by malicious actors using transferable adversarial examples.
