Meta Invests in AI Chips, Privacy Tooling
Meta is rolling out four new generations of in-house AI chips through 2027 to manage AI workloads and reduce dependence on third-party silicon, mirroring Apple's silicon strategy. Meta's latest AI tools also aim to flag suspicious messages and impersonation attempts, exemplifying the type of ML-powered security Apple is known for. But Meta also faces a class-action lawsuit over AI smart glasses privacy.
Meta's silicon push is spearheaded by the "Artemis" chip, designed to handle AI inference workloads, with subsequent generations focusing on improved performance and efficiency. This initiative aims to give Meta greater control over its AI infrastructure and reduce reliance on Nvidia and other external chip suppliers. The AI scam protection tools leverage large language models to analyze message content and user behavior, identifying potential phishing attempts and impersonation tactics with increased accuracy. Meta is also exploring federated learning techniques to train these models on decentralized data, enhancing privacy and minimizing data collection. The class-action lawsuit in Kenya alleges that Meta's smart glasses collect and process biometric data without explicit user consent, violating privacy laws and potentially enabling surveillance. The plaintiffs are seeking damages and demanding stricter data protection measures for AI-powered wearable devices.
