Since I found chatbots (ChatGPT); Everyone is starting to look for ways to take advantage of it, be it at work, college or performing daily tasks, but companies are divided on whether employees should use ChatGPT while working, as some companies allow employees to use it. Some companies allow employees to use it. They blocked it and Samsung was one of the first to allow it, but now it's in big trouble because of this decision.
Samsung's semiconductor division allows engineers to use ChatGPT bots to review source code, but employees inadvertently share confidential information while using the bot to help work.
employees share confidential information with ChatGPT bots; For example: the source code of a new software that the company is developing and the data of recording internal meetings related to Samsung devices. As a result, in less than a month, three Samsung employees were recorded leaking sensitive information through ChatGPT incidents.
How did ChatGPt leak Samsung secrets?
The three scenarios in which a bot crew is used are as follows: first case: employees insert the source code of a new program into the bot to find bugs, second scenario: employees insert a series of top-secret test threads to identify bad chips and ask for help in optimization, and the third scenario : The employee uses the artificial intelligence tool (Naver Clova) to convert the recorded company meeting into a document, and then reads it through the ChatGPT bot into a presentation.
These Samsung trade secrets are now in the hands of the bot's developer, OpenAI, as ChatGPT stores user-entered data for ongoing training. There is no doubt that any kind of data breach could be disastrous for the companies involved, especially in the highly competitive semiconductor industry.
What did Samsung do to protect its secret?
Samsung is investigating the circumstances of the three incidents and, if necessary, will fine the managers and employees that led to the company's data breach, even though Samsung allowed employees to use the tool from the start.
Samsung communicated this information to its managers and employees through an internal notice, urging them to use ChatGPT bots with caution. The data is transmitted and stored on external servers, which makes it impossible for the company to retrieve it. "
In addition, Samsung is taking a number of precautions to prevent similar leaks in the future, with the company informing its executives and employees that it will permanently block access to ChatGPT in the event of a similar incident even after implementing security measures. company network.
And in response to what happened, Samsung is developing its own ChatGPT-like AI tool for internal use only, so employees can get quick help without exposing company secrets.
According to a report by the South Korean edition of The Economist, Samsung took immediate action after discovering the bug, such as limiting the download size of each question in the bot to 1024 bytes.
ChatGPT employees use at work:
A recent survey of more than 11,000 people at Fishbowl, a social network for professionals, found that 43% of employees now use tools like ChatGPT at work, indicating a 27% increase in employees using tools like ChatGPT. Fishbowl will resume in early January 2023.
The study also states that 70% of employees who use these tools (a significant percentage) do so secretly at work without their managers knowing. The study included more than 11,700 employees using the Fishbowl app from major companies such as Amazon, Google, Meta, Twitter, IBM, JPMorgan, and others.
On the other hand, Samsung believes that semiconductor systems can change depending on the artificial intelligence created and urges management to deal with these changes, so it organized an event for all executives on the 22nd of this month on "ChatGPT Emergence, Change". Future “Webinar on the topic through Generative AI” (Jung)-Hee Han, Executive Vice President of Samsung, stresses the importance of ChatGPT and the need to further understand generative AI.