Meta is reportedly planning to use an AI-powered system to evaluate the potential harms and privacy risks of up to 90% of updates made to its apps, including Instagram and WhatsApp. According to internal documents reviewed by NPR, this shift would mark a significant change from the largely human-led privacy review process currently in place.
Privacy Reviews Under Federal Oversight
Meta’s privacy reviews stem from a 2012 agreement with the Federal Trade Commission (FTC), which requires the company to assess the privacy implications of product updates before release. Until now, these evaluations have been performed mostly by human experts to identify and mitigate risks.
Under the new system, product teams will fill out a questionnaire detailing their proposed updates. The AI will then analyze the submission and provide an “instant decision” that highlights identified risks and outlines necessary conditions the update must meet to proceed. This approach aims to speed up the update process while maintaining oversight.
Concerns Over Increased Risks
A former Meta executive told NPR that while the AI system allows for faster updates, it also introduces “higher risks.” They warned that relying heavily on AI may reduce the ability to prevent negative consequences before they impact users, as some risks might be overlooked or underestimated.
In response to the report, a Meta spokesperson highlighted the company’s investment of over $8 billion in privacy programs and its commitment to regulatory compliance. They emphasized that the AI system will handle lower-risk decisions with consistency and speed, but complex or novel issues will still undergo human review.
“As risks evolve and our program matures, we enhance our processes to better identify risks, streamline decision-making, and improve people’s experience,” the spokesperson said.
Author’s Opinion
Automating privacy and risk assessments with AI can improve efficiency, but it’s crucial not to sacrifice thoroughness and accountability. Human judgment remains essential for understanding nuanced, context-dependent risks that AI may miss. Companies like Meta must carefully balance the desire for rapid innovation with the responsibility to protect users from potential harm.
