Concerns Over Confidential Information Leakage During Coding
Amazon Also Displays Usage Restriction Pop-up
[Asia Economy Reporter Jeong Hyunjin] Amid the global craze for the AI chatbot 'ChatGPT,' Microsoft (MS), which invested in ChatGPT developer OpenAI, recently issued a 'ChatGPT usage advisory' to its employees. This is due to concerns that confidential information might leak. Not only MS but also Amazon warned against using ChatGPT as a 'coding assistant' and for other purposes.
◆ "Confidential information could be used for training"
According to the US economic media Business Insider on the 1st (local time), a senior engineer under MS's Chief Technology Officer (CTO) recently responded in an internal forum to a question about whether ChatGPT could be used for work by saying, "Do not input confidential information. This information could be used to train future models." There is concern that if employees inadvertently input confidential company information such as internal software code during work, ChatGPT might learn it and provide that information to other users.
MS's caution toward ChatGPT is notable considering the special relationship between MS and OpenAI, the developer of ChatGPT. MS announced on the 23rd of last month that it would invest in OpenAI. Although MS did not disclose the exact investment amount, Bloomberg reported that the investment could total $10 billion (approximately 12.3 trillion KRW) over several years. This was the third partnership following those in 2019 and 2021. MS also revealed plans to apply OpenAI's technology to its own products.
Vincent Conitzer, a professor of computer science and director of the AI research institute at Carnegie Mellon University, said that MS's close relationship with OpenAI could lead to potential conflicts of interest as MS might receive more training data. Professor Conitzer also pointed out that issues regarding accountability should be considered if confidential information is leaked. He said, "Is it the responsibility of the employee who should not share confidential information, the responsibility of OpenAI to use the information carefully, or the responsibility of both parties?" He added, "As one of the key leaders in the tech industry, MS will have a tremendous impact on this topic."
◆ Amazon even has a popup restricting ChatGPT use
Similar concerns have arisen at Amazon. Business Insider reported on the 25th of last month that Amazon warned its employees in December last year not to upload confidential information to ChatGPT. After OpenAI released ChatGPT in November of the same year, employees flooded the company’s internal messenger Slack channels with questions about whether they could use it for work and, if so, to what extent. In response, Amazon’s legal team said, "The information entered could be used as training data for ChatGPT, and we do not want that information to be leaked externally."
According to the report, Amazon has established internal guardrails regarding ChatGPT use. When employees access the ChatGPT website from within the company, a popup appears stating, "You may be accessing an unauthorized third-party service related to Amazon security." However, Business Insider explained that employees know how to bypass this popup. Business Insider also noted that some Amazon employees are already using ChatGPT as a software 'coding assistant,' showing interest in its potential.
Business Insider conveyed the atmosphere by saying, "Privacy concerns seem to be the smallest issue for Amazon employees. They say using chatbots for work increases productivity tenfold and have even expressed willingness to join teams developing similar services."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
