GPT-5 Pro Helps Immunologist Solve 3-Year T Cell Mystery
OpenAI reports that GPT-5 Pro assisted immunologist Derya Unutmaz in resolving a long-standing question about T cell behavior, with potential implications for cancer and autoimmune research.
AI & ML
Frontier models, research results, agents, and developer tooling.
New MosaicLeaks Framework Tests AI Research Agents' Secrecy
ServiceNow's MosaicLeaks, published on Hugging Face, probes whether research agents can keep confidential information secret.
OpenAI Adds Spend Controls and Usage Analytics to ChatGPT Enterprise
New admin tools give enterprise customers clearer visibility into ChatGPT usage and tighter budget controls, addressing cost governance as adoption scales.
Hugging Face Ships huggingface_hub Weekly With AI and Human Review
The company says its weekly release cadence for the huggingface_hub library combines AI tools, open components, and human oversight.
Four Patterns for Building Cybersecurity Evals
A practical framework for designing security evaluations of AI systems centers on sandboxing, variable input difficulty, and a structured grading process.
Beyond LoRA: Can Other Fine-Tuning Methods Beat It?
Hugging Face asks whether any alternative can outperform LoRA, the most popular parameter-efficient fine-tuning technique.
GLM-5.2 marks a step change for open agentic AI
A new open model crosses a capability threshold that one AI researcher says he has been tracking closely.
Hugging Face Trials Cross-Origin Storage for Transformers.js
An experiment with a proposed browser API could change how Transformers.js stores and loads model assets across origins.
Omio Uses OpenAI to Power Conversational Travel and AI-Native Development
A travel platform’s shift to AI-native development shows how large language models can reshape both customer experience and product building.
Samsung Electronics Rolls Out ChatGPT Enterprise and Codex Globally
The deployment makes Samsung one of OpenAI’s largest enterprise customers, bringing AI chat and coding tools to a global workforce.
OpenAI Backs Shared Standards for Advanced AI via Appia Foundation
OpenAI is contributing to the Appia Foundation's work on evaluation frameworks and safety practices, signaling a push for industry-wide cooperation on advanced AI governance.
Open Source AI Ban Would Be a Mistake, Op-Ed Argues
Nathan Lambert and Kevin Xu argue against restricting open source AI in an op-ed aimed at a general audience.
Import AI 462: Superpersuasion, Self-Sustaining AI, and ASI Paths
Jack Clark's latest newsletter examines persuasive AI, self-sustaining systems, and the road to artificial superintelligence.
Goodfire's Cammarata: Neural Networks Are Already Clean
A new essay argues that agentic analysis reveals most neurons have a single role, potentially simplifying interpretability and alignment.
Hugging Face Unveils Beam, a 501B Open Model
The release signals a push toward large-scale open-weight models that developers can own and adapt.
Stack Overflow for Agents plugin triples signups, says OpenAI
A small team used Codex to build a ChatGPT plugin that dramatically boosted adoption of Stack Overflow's agent platform.
OpenAI reasoning model yields 18 new diagnoses in rare childhood diseases
Researchers used an OpenAI reasoning model to identify 18 new diagnoses in previously unsolved pediatric rare disease cases.
Interconnects Marks Three Years of Weekly AI Writing
Nathan Lambert's mid-2026 update reflects on the durability of independent AI commentary.