Google has recently updated its own privacy rules to clarify the possibility of using i public user data for training AI models (AI), as used by Bard. A transparency in the use of data that could dbecome mandatory with the entry into force of the AI Act. Although currently the Google chatbot is not yet accessible in Europe due to privacy issues.
Google updates privacy policies, uses user data for AI
OpenAI clarified that it collects data from the Internet to train ChatGPT. Google has also now confirmed that it does this data collection activity for ensure transparency and compliance with applicable laws. From July 1, in the “Publicly accessible sources” section of the privacy policy, the company has specified that it will collect public data to train Google AI models such as Google Translate, but also “Bard and AI Cloud functionality”.
Google has therefore specified that public user data can be used for training AI models. By public data we mean not only those available on Google services, but also those shared online in general. What you write on social networks with a public account or on forums and blogs can become “food” for artificial intelligence.
However, some sites are trying to block scraping to train AI. Twitter, for example, has imposed limits on reading tweets to prevent data scraping by third parties. Even if only “temporarily”.
For users, this change of policy does not bring any revolution. But it brings a little more transparency even without the need for the intervention of the AI Act by the European Union.