摘要： 目前,人机共驾接管失败致死的案件备受瞩目,对于安全员虽未履行接管义务,但可否主张信赖原则而不承担过失责任,存在肯定论与否定论的争议。从肯定论看来,人机信任建立在人工智能可信任性的基础上——机器有自主性、能动性和以人类为中心的意向性,并偏重人类对机器的信任。对高度分工、高风险的人机信任适用信赖原则,是组织模式的一种延展,不会导致责任稀薄,更符合刑事责任性质,具备合理性。信赖原则例外否定过失责任时,以不同自动驾驶水平下安全员、自动驾驶系统和人机交互作为客观因素,可具体判断信赖的相当性。安全员没有履行接管义务的行为具有客观违法性,但当其合理信任自动驾驶系统能够执行预期功能时,不能具体预见结果的发生,由此信赖原则在责任阶段否定过失,发挥后端风险结果分配的作用。

关键词: 接管义务, 人机信任, 信赖原则, 过失责任

Abstract: Currently, the case of death caused by a failed takeover by a human-machine co-driver is in the spotlight, and there is a controversy between affirmative and negative theories as to whether a safety officer can claim the principle of reliance and not be held liable for negligence even though he or she failed to fulfill the duty to take over.From the affirmative view, human-machine trust is built on the basis that AI trustworthiness-machines have autonomy, agency, and human-centered intentionality, and favors human trust in machines.The application of the principle of reliance to the highly compartmentalized, high-risk human-machine trust is an extension of the organizational model that does not lead to dilution of responsibility, but is more in line with the nature of criminal responsibility, and possesses reasonableness.When the exception to the principle of reliance negates negligence liability, the equivalence of reliance can be specifically judged by using the safety officer, the autopilot system, and the human-machine interaction at different levels of autopilot as objective factors.The safety officer's behavior of failing to fulfill the obligation to take over has an objective illegality, but when he reasonably trusts that the automatic driving system can perform the expected function, he cannot specifically foresee the occurrence of the result, thus the principle of reliance negates negligence in the stage of liability, and plays a role in the allocation of the results of the back-end risk.

Key words: The Obligation to Take Over, Human-machine Co-driving, Principle of Reliance, Negligence Liability