Corporate Accountability Fractures as AI Agents Take the Reins

The survey of over 1,000 executives and managers highlights a growing vacuum in corporate policy. Roughly 25% of respondents assign fault to the employee utilizing the AI, while an equal portion shifts the burden to that employee’s supervisor. The remaining blame is scattered among risk teams, the AI vendors themselves, or simply left in a state of uncertainty. This lack of consensus persists even as firms like OpenAI, Google, and Meta roll out autonomous agents capable of managing emails, pricing products, and executing multi-step tasks across proprietary systems.

The real-world fallout is already mounting. Leeron Walter, a vice president at Teramind, points to a major U.S. retailer that saw 18,000 customers receive erroneous refund notices in just six hours due to a flawed AI prompt. In another instance, a consumer electronics company mistakenly listed a flagship device at an 85% discount, triggering 700 orders before the error was caught. As tools like Google’s Gemini and Meta’s Muse gain traction, the shift from static chatbots to active, credentialed agents makes these high-stakes errors increasingly difficult to contain or attribute.