Nefarious actors can use open models maliciously, [...] whereas developers of closed models have more means to identify and disrupt this malicious use. Unverified source (2024)