Meta plans to automate many of its product risk assessments

According to internal documents, an AI-powered system soon may take responsibility for evaluating possible losses and privacy risks on 90% updates on meta applications like Instagram and WhatsApp Been seen by NPRThe

NPR says A 2012 contract It is necessary to handle the privacy review of its products of its products by evaluating the risks of any possible updates between Facebook (now Meta) and the Federal Trade Commission. To now, these reviews are mostly run by human evaluator.

Under the new system, Meta reports that the product teams will be asked to fill a questionnaire about their work, then usually get an “instant decision” with the AI-marked risk, as well as any updates or features with the requirements.

This AI-centric system will allow the meta to update its products faster, but a former executive told the NPR that it also created a higher risk, because the negative externalities of the product change are less likely to prevent them before creating world problems. “

In a statement, Meta confirmed that it was changing its review system, but it emphasized that only “short-risky decisions” would be automated, while on the other hand the “human skills” would still be used to test “novels” novels and complex issues.