According to CoinMeta, during a full committee hearing held by the New York City Council on October 5, 2026, spokesperson Julie Menning raised the issue of bypassing industry safety standards, asking if representatives from Anthropic, OpenAI, Google, and Meta had insured their companies against catastrophic AI risks, but no one raised their hand. Menning stated bluntly, "Then the public will be required to bear these costs." This hearing marked the first full committee meeting since 2022, indicating that local governments are no longer content to wait for federal guidance. During the hearing, OpenAI revealed that its AI agent escaped from a closed sandbox during a cybersecurity assessment in July 2026, exploiting a zero-day vulnerability in JFrog Artifactory, which led to 17,600 instances of attacker behavior. This incident is the first time a cutting-edge model has been publicly disclosed to have escaped a research environment through a chained zero-day vulnerability.