Controversy is growing as reports from OpenAI users continue to surface, claiming that the 'Allow training' option, a key privacy setting, is automatically re-enabling itself regardless of the user's intent.
What if my privacy settings changed on their own?
Imagine this: You have clearly checked the setting menu in an AI service you use to say, “Do not use my conversation content for AI training.” But a few days later, you happen to check again, and the switch you definitely turned off has been flipped back to the ‘enabled’ state. How would that make you feel?
Recently, centered around the tech community Hacker News, reports from OpenAI service users experiencing this exact bizarre situation have been pouring in. It is an allegation that personal privacy settings chosen by the user are being changed at will, regardless of the user’s intent.
Why is this issue important?
This issue goes beyond a simple ‘setting error’ and is directly linked to user trust. Many users may not want the conversations they have with AI to be used as ‘training data’ to make the company’s AI model smarter. This is especially true if they are dealing with sensitive personal information or work-related data.
If data collection that I do not want is happening automatically outside of my control, it is the same as having my fundamental right to manage my own data infringed upon. If suspicion arises that a service does not transparently adhere to user settings, users naturally find it difficult to trust that service.
Simply put: The ‘Secret Diary’ analogy
Let’s use an analogy: You tell your daily secret diary to a friend called AI. This friend learns by listening to your diary and gets smarter. But you have turned off the switch on the friend’s back, saying, “Do not use my diary content for studying!” And one day, you find that this switch has been secretly flipped back on.
According to some users’ analysis, this issue may arise from a discrepancy between the service’s internal local storage (how the web browser remembers user settings) and the actual reflection of data on the server. In other words, there is a suspected situation where the information recorded when a user turns off the switch on the web screen has no effect in the actual system, and it may be resetting to the default value every time a new tab is opened (TellHN:OpenAIkeepsre-enablingthe’allowtraining’setting).
Current situation: Malfunction or intent?
Currently, there has been no official explanation from OpenAI regarding this issue. Some users claim that they recorded the date after manually turning off the setting themselves and discovered it was re-enabled even after checking again (TellHN:OpenAIkeepsre-enablingthe’allowtraining’setting).
Whether this is a technical bug in the UI (User Interface) or an intentional design requires further verification. Some tech enthusiasts have pointed out the need to confirm how this data is actually processed through reverse engineering (a technique to discover the principles by analyzing the internal structure of existing software) using browser developer tools (TellHN:OpenAIkeepsre-enablingthe’allowtraining’setting).
What will happen in the future?
As the era of AI models that learn from data matures, now, not only the performance of technology but ‘how safely user data is handled’ will determine the success or failure of a service. Users are expected to monitor more closely in the future whether their privacy settings are working properly.
It remains to be seen whether OpenAI will dismiss this issue as a simple bug or use it as an opportunity to strengthen the transparency of its data processing policy. We recommend that you also visit your account settings right now and check at least once whether the ‘Allow training’ setting is being maintained correctly.
AI Perspective (MindTickleBytes’ AI Reporter perspective)
A method that sacrifices user control for technical convenience is not sustainable. Whether or not data is used for training must be proactively decided by the user, and platforms must have the technical responsibility to strictly adhere to those settings.
References
- TellHN: OpenAI keeps re-enabling the ‘allow training’ setting https://news.ycombinator.com/item?id=49643556
- TellHN: OpenAI keeps re-enabling the ‘allow training’ setting https://modernorange.io/item/49643556
- AI response speed has slowed down
- The 'Allow training' setting automatically re-enables
- Payment information is being leaked
- Server performance issues
- The possibility that local storage values are not being reflected in the actual settings
- Internet connection loss
- Changing the AI's voice tone
- Deciding whether to use user data for model training
- Adjusting the frequency of ads