Some workers say the issues could ‘potentially induce cardiac incidents in Legal.’
At its annual Amazon Web Services developer conference, Amazon announced Q, the company’s AI chatbot for business. As it turns out, Q might not be quite ready for primetime.
According to leaked documents viewed by tech newsletter Platformer, Q is “experiencing severe hallucinations and leaking confidential data.” Some of those leaks include the location of AWS data centers, internal discount programs, and even unreleased features.
In a statement to Platformer, an Amazon representative refuted claims that the chatbot is leaking confidential information.
“Some employees are sharing feedback through internal channels and ticketing systems, which is standard practice at Amazon,” a spokesperson said. “No security issue was identified as a result of that feedback. We appreciate all the feedback we’ve received and will continue to tune Q as it transitions from being a product in preview to being generally available.”
Q is a business-focused chatbot that can be customized to assist employees with a variety of different job functions, as well as answer questions and generate content.
When it announced the product, Amazon said several companies already use Q, including Accenture, Amazon, BMW Group, Gilead, Mission Cloud, Orbit Irrigation, and Wunderkind.
Rather than pull information from the web as many other chatbots do, Q pulls answers from the “company’s information repositories, code bases, and enterprise systems,” making it even more problematic if that information is incorrect or shared with the wrong people.
In November, OpenAI debuted its customizable chatbots, GPTs, that can answer questions using in-house data rather than the web. However, the rollout of its GPT Store has been delayed due to the CEO-related turmoil.
