Although it took longer than expected to arrive, Bard has been available here for a few months, once Google made sure that its long-awaited chatbot was properly adapted to European regulations. Something that seems to me the most sensible thing to do given the formidable level of scrutiny all its operations are subject to, even if it has delayed its arrival in the common European space by about two months from its main deployment in Smět.

As I indicated, Google has ensured the adaptation of the service to the regulations of the European Union, with modifications focused mainly on information visibility and privacy and data protection functions. In this way, we have detailed information about what is happening with our data and the queries we make on the service, and we are also informed about the control we can have over this data.

The common denominator of chatbots, not just Bard, is that Our questions and answers can be used by the companies that offer them, in this case Google. The reason explanation is usually a general “we use your data to improve the service”, which can mean anything from using it to retrain a model to analyzing it for errors in the answers, information that can be used to refine the model’s operation. In all cases, it is clear that we should not think that this information will be lost like tears in the rain once we end the conversation.

However, it seems that despite the warnings, there are still many users who are not aware of this, and that much Google has updated the Bard Privacy Center to remind users not to provide personal information to the chatbot. In it, and highlighted in bold, we find the following:

«Do not enter confidential information or any data in your conversations with Bard that you would not want the reviewer to see or for Google to use to improve our products, services and machine learning technologies..»

This means that if we enter personal/private data, it can be used for the aforementioned improvement of the service, which means that even may be displayed by human reviewers, which analyze some interviews. In this regard, Google indicates that they are taking proactive measures to protect user privacy, but even they have the possibility that some of them will be seen or, worse, seen by a model. , which could later reproduce it. It’s highly unlikely, yes, but not impossible.