engblogs

OpenAI defends user privacy against a New York Times court demand by outlining a privacy-first security roadmap, including de-identification, client-side encryption, secure data handling, and legal safeguards to protect conversations.

Introduction to GPT-5.1 Instant and GPT-5.1 Thinking models featuring enhanced conversational abilities, adaptive reasoning, precise thinking time adjustments, and updated safety evaluations including mental health and emotional reliance considerations.

GPT-5.1 introduces smarter, more conversational AI models with enhanced adaptive reasoning, improved instruction following, and customizable tone settings for a personalized ChatGPT experience.

OpenAI challenges The New York Times’ demand to hand over millions of private ChatGPT conversations, emphasizing its commitment to user privacy, legal defenses, and enhanced security measures including client-side encryption and data de-identification.

Neuro leverages ChatGPT Business to enhance efficiency, drive retail success, and innovate across marketing, legal, and customer insights while scaling nationally.

Stanford’s Hazy Research Lab introduces Intelligence per Watt (IPW), a novel metric optimizing AI efficiency by measuring accuracy relative to power consumption, promoting edge computing and smarter AI deployment with support from Snorkel AI's domain-specific benchmarks.

Lambda’s MLPerf Training v5.1 results demonstrate the NVIDIA GB300 NVL72 cluster delivering up to 27% faster LLM training performance than the previous GB200 NVL72, leveraging NVIDIA Blackwell Ultra architecture and optimized software stacks for superior AI training efficiency.

Explore how Cloudflare's remote bindings architecture enables seamless local development by connecting Workers to real production resources like KV, R2, and D1, improving testing accuracy and developer experience.

Explore how Stripe Billing’s latest enhancements enable businesses to streamline multiprocessor payments, implement flexible and hybrid pricing models, and optimize AI product monetization with innovative tools like the LLM proxy.

Exploring the computational foundations and emerging commonalities of human-like intelligence in AI through research in computer vision, machine learning, and representation learning.

A novel alignment method reorganizes AI visual representations to mirror human conceptual hierarchies, enhancing robustness, reliability, and task performance in vision models.

OpenAI offers a free year of ChatGPT Plus to U.S. servicemembers and veterans transitioning to civilian life, providing AI-powered assistance for resumes, interviews, benefits, and planning.