Facebook spying on users by reading and analyzing private posts and conversations? This is what emerges from a new report published by Reuters
That there was a question-privacy related to the galaxy of Facebook services had long been clear. Mark Zuckerberg’s social network – and the group’s other platforms such as Instagram and WhatsApp – has repeatedly come under the spotlight for issues related to the management of user data. A new report published by Reuters casts a further shadow on this front.
The document states that a select team of auditors continuously reads and analyzes users’ posts. The goal is to help systems improve services and features related to Artificial Intelligence. Instagram is also named in the report, so the company’s employees would also read the conversations present in this social network of Zuckerberg’s house. There would even be a reading method adopted by the company in order to better analyze content and chats, and it would also record users’ names and other sensitive information.
How does Facebook’s post analysis work?
The Reuters report confirms a widespread fear: Facebook monitors what users say, not only in public posts but also in private conversations. According to the paper, reviewers first analyze content based on five parameters: content (selfie, panoramic photo, text…), occasion (birthday, event…), expressive elements (opinions and feelings), intent (plan, inspire, motivate, entertain…) and setting (professional, informal, scurrilous…).
Content is then captured and archived based on its type. Comments are also subjected to this treatment, and are accompanied by the name and sensitive data of the author. To analyze the posts, Facebook relies on the outsourcing collaboration of an Indian company called Wipro. There would be more than 260 employees who daily annotate and analyze Facebook and Instagram posts.
The goal is to better understand the content uploaded by the platforms in order to achieve increasingly sophisticated machine learning, and thus meet the needs of corporate clients. AI algorithms currently cannot understand the complexity of human language, but with the accumulation of Big Data, they want to increase their level of understanding.
The goal becomes difficult to achieve in platforms such as Facebook and Instagram where there is a succession of different types of content (text, photos, videos, links…) and where there are millions of different languages, as well as slang, idioms and neologisms.
What are the dangers to privacy?
The company therefore wants to train its AI systems through careful analysis of public posts. This would improve the machine learning algorithms and services that the platform offers to businesses and private users. So this operation would not be totally negative.
The problem, however, is related to transparency and privacy policy that would have no reference to this process operated by Mark Zuckerberg’s group. So users are completely unaware of the existence of these algorithms and the work done. So they don’t even have the option to reject the post tagging system, not to mention that these could be misused by Facebook employees. In short, the company, already involved in several scandals related to illegitimate accumulation and data leakage, continues to exploit conversations for its own purposes without accurately informing users.