ChatGPT’s reply actually leaked corporate secrets. The Ministry of Economic Affairs warned: AI leakage has become a new type of risk.
Quick Look
The Intelligence Bureau of the Ministry of Economic Affairs warned that people and corporate employees may ignore information privacy when using AI tools, and enter business secrets into generative AI such as ChatGPT, which may be leaked when the AI answers questions from other users, affecting the competitive advantage of enterprises.
AI-generated summary
Why It Matters
The Intelligence Bureau of the Ministry of Economic Affairs issued a warning regarding the development of artificial intelligence and corporate information security, pointing out that the generative AI training mechanism may lead to the leakage of secrets.
The Intelligence Bureau of the Ministry of Economic Affairs warns of the risk of AI leaks. (Reuters)
[Reporter Liao Jianing/Taipei Report] While people rely on the convenience of AI (artificial intelligence), they may neglect the protection of information privacy. AI is training and strengthening its own functions through a large amount of dialogue data provided by all walks of life. In the process, there is the risk of information privacy and business secrets being leaked. The Intelligence Bureau of the Ministry of Economic Affairs has issued a special warning. In one case, an employee of Company A entered the company's business secrets into ChatGPT, and an employee of Company B asked a similar question to ChatGPT. The AI system may use the content of the business secrets entered by the employees of Company A to generate answers to the employees of Company B.
The Wisdom Bureau pointed out that a company’s business secrets are known as the “crown jewels.” Technology or know-how that companies invest a lot of research and development costs has industrial competitive advantages and is an important asset of the company. Once leaked, the company’s research and development efforts will be in vain, and it will face being overtaken by competitors’ technology.
The Wisdom Bureau pointed out that common leaks include employees reproducing business secret files through emails, flash drives, computer screenshots, or taking photos with mobile phones, or making paper copies and bringing them out or providing them to the new company for use. Or employees install software on official computers that is not authorized by the company, use the software to download the company's business secret files, upload them to personal hard drives and take them out of the company for use, etc., or employees from cooperating manufacturers take advantage of the opportunity to enter other companies to maintain equipment and copy the process parameters of the machines.
In addition to common leaks, the Wisdom Bureau specifically warns about AI leaks. Judging from the revelation of the South Korean Samsung company's use of generative AI leaks in April 2023, company employee A entered a company's business secret into ChatGPT.
Open Questions
- How can companies effectively prevent employees from misusing AI to leak secrets?
- Will the government legislate laws to regulate companies’ use of AI tools?
