From Meta Engineering · 15 stories
Strategic Frameworks and Systems Vital for Successful AI Integration in Enterprises
AI's integration in enterprises is moving beyond model development to focus on creating robust systems for execution and governance. This shift highlights the importance of developing adaptable frameworks to support AI's role across various functions such as finance, HR, and operations. It reflects a broader industry trend where the focus is on building the necessary infrastructure to ensure AI's ongoing, safe, and productive incorporation into real-world workflows, addressing the current challenges and limitations.
Meta updates smart glasses to disable camera if privacy light tampered
Meta is releasing a mandatory update for its smart glasses that disables the camera if the privacy LED is tampered with. This initiative addresses privacy concerns and misuse incidents involving covert recording. The update coincides with reports of new Meta glasses that may continuously record without a privacy light, raising further privacy issues.
New AI Models for Long-Horizon Coding Tasks Introduced
Several AI models aimed at long-horizon tasks in coding and robotics have been released. GLM-5.2 by Hugging Face extends support for coding-agent scenarios with a 1 million token context. Cognition's SWE-1.7 enhances long-horizon asynchronous tasks with reinforcement learning. Xiaomi-Robotics-1 combines vast pre-training data for improved robotics capabilities. These releases highlight advances in scaling and reasoning capabilities.
Meta Introduces Hybrid Asset Classification for Privacy-Aware Infrastructure
Meta has unveiled a hybrid asset classification strategy using large language models (LLMs) to handle ambiguous data in privacy-aware infrastructure while maintaining deterministic rules for enforcement. This method addresses the complexities of AI-native products with varied data inputs, ensuring compliance and effective data governance. It is a response to the challenges posed by the increasing speed and scale of AI innovations, and the approach aims to better manage privacy controls for evolving AI products.
Meta Announces Petal, a Petabit-Class Transoceanic Subsea Cable Using Multi-Core Fiber
Meta is building Petal, a new subsea cable connecting France and the United States, which will be the first to deliver petabit capacity over transoceanic distances. This project will deploy multi-core fiber technology at scale, doubling capacity per fiber without a proportional increase in power or physical infrastructure, addressing increasing demand for global connectivity.
WhatsApp Rolls Out Optional On-Device Scam Alert Feature in Limited Beta
WhatsApp has launched a limited beta of "Scam Alert," an optional feature that uses an on-device machine learning model to identify suspicious messages from unknown contacts. This feature processes message content locally on the user's device to maintain end-to-end encryption while alerting users to potential scams. Users can block, report, or ignore flagged messages, and can opt to share message data to improve the model's accuracy.
SilverTorch Introduced as a Unified Architecture for Recommendation Systems
SilverTorch is a new retrieval paradigm for recommendation systems that integrates various components into a single neural network. It demonstrates a significant increase in throughput and compute efficiency compared to traditional microservice-based systems, enhancing the quality and speed of recommendations.
Meta Open-Sources Rebalancer, a Tool for Solving Assignment Problems
Meta has open-sourced Rebalancer, a library used internally for nine years to solve resource allocation and assignment problems across its infrastructure. This tool helps optimize object-to-bin assignments, such as hardware, service, and task placements, by separating problem specification from its solution to improve usability and scalability.
Meta Ads Introduces Hierarchical Interest Representation for Ad Optimization
Meta has launched Hierarchical Interest Representation to enhance ad deep funnel optimization. This system uses advanced graph learning to connect user interests with advertiser products, improving ad relevance and engagement.
Meta Implements Open-Source Kernel Scheduler to Improve Ads Performance
Meta upgraded its ad service latency management by utilizing the open-source sched_ext framework. This change resulted in a 28% reduction in retrieval latency, improved ad ranking, and reduced energy consumption.
Meta Marks 10 Years of Support for Python Software Foundation
Meta celebrates a decade of sponsoring the Python Software Foundation, emphasizing Python’s significance in its infrastructure and AI research. This commitment aims to promote the language's growth and inspire additional support from others in the tech community.
Meta Develops Ultra-Narrow Steel-Can Batteries for Smart Glasses
Meta has engineered new steel-can batteries for smart glasses, achieving widths of 7mm. This innovation allows for better energy efficiency and performance in tight spaces, addressing limitations of traditional battery types.
Meta Adopts AV1 Codec for Enhanced Real-Time Communication
Meta has implemented the AV1 video codec for real-time communication, enhancing video call quality while reducing bandwidth use by at least 20% compared to H.264/AVC. This upgrade improves the experience for users on slower networks, particularly in emerging markets.
Meta Launches Instantaneous PowerLoss Storm for Data Center Disaster Readiness
Meta has introduced the Instantaneous PowerLoss Storm, a new testing paradigm to assess data center readiness for sudden power losses. This initiative is critical as it enhances Meta's infrastructure resilience against zero-notice disasters, ensuring minimal impact on fleet availability.
Meta Introduces ZGateway Proxy for ZippyDB to Unify Traffic and Improve Management
Meta has introduced ZGateway, a new proxy layer for ZippyDB, its widely-used key-value store, to unify client traffic and manage its extensive client fleet. This proxy enables capabilities like admission control, load balancing, and cross-region resilience, which were difficult to implement directly in individual client binaries.