A petition filed with the Supreme Court included dozens of non-existent legal citations and court rulings. In that case, Justice Gila Canfy-Steinitz took the opportunity to convey a message extending beyond the specific case. According to her, even if the sources were fabricated by AI, the responsibility to verify their authenticity rests on the professional using them. In other words, the Supreme Court ruled that claiming AI is responsible for an error cannot serve as a "line of defense" against professional liability.
Although the case involved a female lawyer, its implications are far broader. According to a survey by the Bank of Israel and the Israel Democracy Institute from May 2026, 57% of workers in Israel already use artificial intelligence tools in their work at least once a week. Programmers, academics, lawyers, accountants, marketing professionals, engineers, and human resources employees routinely use a variety of AI tools to write code, draft documents, analyze information, and generate output quickly. The use of AI has transformed in a short time from a technological novelty into a routine work tool.
Who is responsible when AI gets it wrong?
However, as its use expands, a question that labor law has yet to answer becomes sharper: When an employee relies on AI and a mistake occurs, who is responsible? Is it the negligence of the employee, who was required to verify the outcome, or does the employer also share responsibility if they encouraged their employees to use these tools but failed to set clear work rules for their usage?
Israeli law still does not provide an explicit answer. An employee's duties of care, loyalty, and professionalism, anchored in law, were designed in a world where decisions were made by humans. Today, some tasks are carried out using systems that provide fast and persuasive answers, but also ones that may be incorrect. Furthermore, the use of AI can sometimes infringe on client privacy and expose their personal information. As AI becomes integrated into the core of professional work, the boundary between human error and technological error becomes blurred.
Such errors can also have immediate employment consequences: Damage to trust, disciplinary proceedings, and even dismissal. Therefore, the distinction between an individual professional error and an organizational failure in implementing the tool is not a theoretical question, but one that could determine who will pay the price.
Why employers may share the responsibility
An OECD study found that employees using AI report an improvement in work productivity, but warned against over-reliance on system outputs without control mechanisms. Large international organizations are beginning to adopt internal policies and guidelines for AI use. Among other things, they prohibit entering confidential information into open systems, mandate human verification of outputs, and clarify that ultimate responsibility remains with the employee. In Israel, by contrast, most workplaces still lack a uniform policy, and employees adopt the new tools, sometimes without organized guidelines.
At Microsoft, for example, there is a clear directive stipulating that all employees and contractor workers are required to undergo company-wide security training, including the safe use of generative artificial intelligence, which includes guidance on which AI tools to use and under what circumstances. The company also provides employees with a portal featuring guidelines for working with language models and adapting tools to various tasks. "Even though the company's guidelines do not address whether the employer or the employee is responsible if a mistake is made, they demonstrate that employers are aware that such errors can occur and that they bear a certain responsibility in preventing them, including through guidelines and training specifically related to this matter. However, even these guidelines do not clearly establish how responsibility will be divided and what will happen if AI generates an incorrect professional output."
It seems difficult to assume that responsibility will continue to rest exclusively on the employee. As employers benefit from the increased productivity and time savings enabled by artificial intelligence, the expectation may also grow that they establish clear usage rules and define which tasks can rely on AI and which require human verification.
It is possible that in the future, just as employers are required to train employees on safety, information security, or the prevention of sexual harassment, the question will also arise as to whether they must train employees in the prudent use of artificial intelligence. If that indeed happens, responsibility for AI usage will no longer lie solely with the employee, but also with the workplace.
The author is an attorney at Gideon Rubin & Co. specializing in labor law.