Data Privacy Happenings
Hello and welcome to the second edition of Mine's monthly newsletter, The Privacy Mindset! 👋
In between all the awestruck reactions and explosion in self-styled AI experts in the wake of OpenAI's release of ChatGPT, data privacy concerns have failed to emerge as a consensus worry. While some have pointed out concerns and others have exposed the model as one that frequently invents sources and presents misleading information, the prevailing narrative has been that we are at a turning point in technological history.
For anyone with an inkling of interest in or knowledge of data privacy, hopefully that is not the case. As if the above problems were not enough, Microsoft, one of OpenAI's main investors, has announced the revolutionary upstart will use its AI to power Bing.
To say Bing's AI has had a tough time of things since the announcement is an understatement. Misinformation? Check. Insulting users? Check. Users messing with the chatbot to crack it? Check.
None of that would be worrisome if these chatbots had not vacuumed up virtually the entire internet and all the data on it before launching, or if they were not storing and processing terabytes of information from all its user conversations every few minutes.
But they are.
OpenAI was not granted explicit access to go through all of that data, some of it surely yours and mine, nor does it seem like OpenAI considers itself bound to any data regulations considering it was built on the very opposite of data minimization.
There is currently no way for individuals to exercise their data rights and inquire as to what ChatGPT has stored, largely because that data is not merely sitting in data centers, but feeding into the system's evolution itself, making it incredibly difficult to isolate and extract.
Few can be sure of the specifics of how OpenAI and Microsoft are handling and processing the data ChatGPT receives, but from what we can infer about its interactions with users, these chatbots are playing with fire when it comes to data protection.
Given many people's pursuits to break these systems, as well as the systems themselves already mixing up users when prompted to recall past conversations, any leak or outage could result in a data privacy catastrophe with tons of data--sensitive or not--being exposed to the public.
That is a step backward for the internet and one we collectively need to consider more as we integrate this AI technology into our work and life.
