An AI-driven system can soon take responsibility for evaluating potential damage and privacy risks of up to 90% of the updates made for meta apps such as Instagram and WhatsApp, according to internal documents Allegedly viewed by NPR.
NPR says An agreement of 2012 Between Facebook (now Meta) and the Federal Trade Commission requires that the company carries out privacy assessments of its products and evaluates the risks of possible updates. So far, those reviews have been largely performed by human evaluators.
Under the new system, Meta reportedly said that product teams will be asked to complete a questionnaire about their work, and then usually receives an “immediate decision” with AI-Geidified risks, along with requirements that an update or function must meet before it is launched.
This AI-centric approach would enable Meta to update its products faster, but a former director told NPR that it also creates ‘higher risks’, because ‘negative external effects of product changes are less likely before they start causing problems in the world.’
In a statement, Meta seemed to confirm that it changed its assessment system, but it insisted that only “decisions with a low risk” will be automated, while “human expertise” will still be used to investigate “new and complex issues”.
#Meta #planning #automate #product #risks #Techcrunch