engblogs

An amended Microsoft-OpenAI agreement delivers long-term clarity and accelerates a scalable AI partnership, with Azure as the primary cloud, cross-cloud product shipping under a non-exclusive OpenAI IP license through 2032, and revised revenue terms to support AI platform growth.

OpenAI secures FedRAMP 20x Moderate authorization for ChatGPT Enterprise and API Platform, delivering secure, governed access to frontier AI for federal agencies with automated evidence, reusable authorization data, and cloud-native security.

Pinterest shares a unified, single-head, multi-task shopping conversion candidate generation model trained across multiple surfaces with a parallel DCN v2 and MLP cross layers, augmented by advertiser-level objectives to boost conversions, recall, CTR, and RoAS.

Design and deployment of a unified multi-task shopping conversion candidate generation model at Pinterest, featuring a parallel DCN v2 and MLP cross-architecture across multiple surfaces to boost conversions, recall, and RoAS while mitigating sparse, noisy offsite signals.

Google DeepMind announces a Korea partnership to accelerate scientific breakthroughs through frontier AI, establishing an AI Campus in Seoul and collaborating with MSIT on a national AI strategy across life sciences, energy, weather and climate.

Five guiding principles to democratize AGI, empower individuals, and pursue universal prosperity through resilient, adaptable, and safely deployed technology.

Principled blueprint for democratizing AI: empowering individuals, pursuing universal prosperity, and building resilient, adaptable systems through safe, collaborative governance and iterative deployment.

A principled framework for responsible AI development, emphasizing democratization, empowerment, universal prosperity, resilience, and adaptability to maximize societal benefit.

A browser-based, real-time collaborative workspace that blends local transcription, Automerge-backed shared documents, and LLM-driven tool generation to turn chat, calls, and a canvas into a living, design-forward environment.

A technical exploration of OpenAI's AI Jobs Transition Framework, mapping 921 occupations into four trajectories: high automation risk, reorganization, growth with AI, and less-immediate change, and outlining policy, training, and real-time workforce implications.

An analysis of how AI reshapes employment across 921 occupations using OpenAI's AI Jobs Transition Framework, detailing task share, human oversight, demand effects, and policy implications from high automation risk to growth.

Using OpenAI's AI Jobs Transition Framework, this post analyzes how AI reshapes occupations—from high-automation risks to growth areas—through task exposure, human oversight, and demand effects, with policy and measurement implications.