Google updates privacy policy to allow AI models to be trained using public information
The search engine giant Google recently updated its privacy policy to use publicly available information on the Internet to train AI models.
The policy mentions that Google can now use publicly available information on the Internet or from other public sources to help train Google’s AI models and build useful features such as Google Translate, Bard (Google’s chatbot, similar to ChatGPT and Microsoft Bing Chat) and Google Cloud AI. In addition, Google has changed the original “language model” to “AI model”.
This means that Google will be able to use information publicly available on the Internet to build a more complete version of its own AI products, rather than the single feature it used to have.
As of now, however, the Chinese version of the privacy policy has not been adjusted accordingly.
Engadget commented that “Google will make it clear to people that any information they post publicly online can be used to train Bard, as well as any other generative AI products Google develops.”
The AIGC system, including ChatGPT, has long been subject to a great deal of copyright and privacy controversy as it trains content based on data available on the Internet.
Two authors sued OpenAI in San Francisco federal court, arguing that OpenAI misused their work to train ChatGPT with their work. the complaint estimates that OpenAI’s training data contains at least 300,000 books, many of which are copyrighted books that were obtained without permission and illegally.
