AI researcher Zico Kolter has demonstrated that OpenAI has not yet patched a jailbreak in ChatGPT, discovered a year ago, which allows it to assist in committing crimes such as hotwiring a car. Additionally, a hacker last year gained access to OpenAI's internal messaging systems and stole details about the design of the company's AI technologies. This security breach, reported by Cade Metz, has raised concerns that foreign adversaries, including China, could potentially exploit similar vulnerabilities.