AI TRENDS | Anthropic Rejects Recall Calls Over Alleged Jailbreak Vulnerability

Anthropic said it disagreed that the existence of a potential jailbreak vulnerability should be grounds to recall commercial AI models that have already been deployed to hundreds of millions of users. According to Jin10, the company argued that a claimed or possible jailbreak issue, by itself, does not justify pulling back models that are already in commercial use. Anthropic did not provide additional details in the statement, including the specific model involved, the nature of the alleged vulnerability, or any mitigation measures.