Ever since AI research firm OpenAI released ChatGPT in November 2022, nothing else has been talked about in the field of technology. In this regard, the e-commerce giant (but not only) Amazon has proved to be attentive and has even warned employees not to interact with ChatGPT and share confidential information.
Amazon prohibits its employees from sharing confidential information with ChatGPT
Amazon's internal Slack channel had seen a lot of employee questions about how to use ChatGPT. Some employees asked Amazon if there were any official guidelines for using ChatGPT on work devices. Others questioned whether they were allowed to work with AI tools. An employee is urging Amazon's cloud computing division, AWS, to clarify his position on using "generative artificial intelligence (AIGC) tools."
After poco, an Amazon corporate lawyer joined the discussion. Screenshots of internal communications in the Slack channel show the lawyer warning employees not to give ChatGPT "any confidential Amazon information," including the Amazon code being written. He also advised employees to follow the company's existing non-disclosure policy, as some of ChatGPT's responses appeared very similar to those Amazon produced internally.
“This is critical because your input can be used as training data for ChatGPT and we don't want its output to contain or look like our confidential information,” the lawyer wrote.
These exchanges suggest that the sudden emergence of ChatGPT has raised many new ethical questions. ChatGPT is a conversational AI tool that answers questions with clearer and smarter answers. The rapid proliferation of ChatGPT has the potential to disrupt several industries, including media, academia, and healthcare, prompting efforts to find new use cases for chatbots and their possible impact.
How employees share sensitive information with ChatGPT and what its developer, OpenAI, does with it could become a thorny issue. This is especially important for Amazon, since archrival Microsoft has invested heavily in OpenAI, including a new funding round this week that is reportedly worth $10 billion.
Emily Bender, who teaches computational linguistics at the University of Washington, said: “OpenAI is far from transparent about how it uses data, but if the data is used for training, I expect companies to think: after several months of widespread use of ChatGPT, is it possible to obtain confidential information of a private company through carefully crafted requests?”
Amazon has a number of internal safeguards for employees using ChatGPT. For example, screenshots from the exchange show that when employees use work devices to access the ChatGPT website, a warning message appears saying they are about to access a third-party service that “may not be approved for use by Amazon Security”.
Employees who participated in the Slack channel chat said they could dismiss the message by simply clicking the "Acknowledge" tab. Staff speculated that the warning popup was to prevent employees from pasting sensitive information to ChatGPT, especially since they hadn't seen the company's policy on internal use.
However, some Amazon employees are already using AI tools as software "coding assistants" to help improve internal lines of code. “I think it would be great to have this feature out of the box right now. So any guide would be great,” an Amazon employee wrote on a Slack channel.
Another employee said they shared Amazon interview questions for an open programming position on ChatGPT. According to the employee's Slack post, the AI model provided the correct solutions to many of the technical problems. “I am both concerned and excited to see what impact this will have on the way we conduct coding interviews,” he said.
Overall, Amazon employees participating in Slack channel chats were excited about the potential of ChatGPT and wondered if Amazon was developing a competing product. The company lawyer who warned employees against using ChatGPT said Amazon is accelerating the development of "similar technology," citing Alexa voice assistant and code recommendation service CodeWhisperer as examples.
An AWS employee wrote that the company's support team recently formed a small internal team to "understand the impact of advanced chat AI on our business." Research shows that ChatGPT is "very good" at answering AWS support questions, including troubleshooting the Aurora database, among others. At the same time, he is also excellent at creating training materials for the AWS Certified Cloud Architect exam, addressing the client's business goals.
The growing use of ChatGPT at work raises serious questions about how OpenAI plans to use shared material with AI tools, said UW's Bender.