OpenAI reaffirms Zero Data Retention for eligible API customers and previews Private Safety Processing, enabling advanced AI safety without compromising data privacy.

A roundup of AI developments: a new arXiv paper on mathematics in the age of AI, a blog explaining Claude's watermark, a Twitter thread on Claude writing a macOS driver, Linear's data on AI usage in software teams, an interactive model architecture tool, a Claude Code pipeline for iOS, a free book on inference engineering, and a hybrid RAG benchmark.

OpenAI announced new safeguards for frontier AI models, including monitoring, alignment, and security measures, to guide development pace amid cyber threats. NVIDIA emphasized securing AI factories, the infrastructure where compute transforms data into intelligence, highlighting the need for full-stack security across chips, networking, and power.

Liquid AI published LFM2.5 Q4_0 checkpoints, produced through quantization-aware distillation (QAD). The method integrates quantization during training to preserve model quality at 4-bit precision. The checkpoints are available on Hugging Face, offering efficient deployment options for edge and resource-constrained environments.

LangChain released langgraph-sdk 0.4.3, adding a decrypt replacement result feature and supporting clearing cron end_time via update(end_time=None). The update includes dependency bumps and internal fixes across the LangGraph ecosystem.

OpenAI announced an initiative to strengthen democratic oversight of AI in national security, providing tools, training, and expertise to government institutions. It also funded 14 independent projects exploring AI policy ideas for economic opportunity and societal resilience, and joined the PORTS-Pike project to support community investment and jobs in Southern Ohio.

OpenAI is expanding ChatGPT Ads to 31 European markets, enabling advertisers to reach users as they explore, compare options, and make decisions within the platform.

IBM Research introduces ALTK-Evolve-HMM, a method to evolve agent memory architectures using hidden Markov models. The approach automatically designs memory modules, improving performance on long-horizon tasks while reducing memory usage. The work is detailed in a Hugging Face blog post.

Google announced partnerships between Gemini, Pixel, and five global football clubs to enhance fan matchday experiences using AI and smartphone technology. The collaboration aims to bring fans closer to the game through innovative features.

Mojo, the Python-inspired language for GPU programming, is now open source. The compiler and toolchain were released under an Apache 2 license, following the 1.0 release last week. Originally planned as a Python superset, Mojo is now its own language, prioritizing GPU programming ease.

Mistral released CLI v0.3.0 with OIDC login/logout/whoami commands and CLI-based app evals. Mistral Vibe v2.24.2 adds a session picker, /log-level, /retry for ACP clients, worktree cleanup, faster startup, and multiple bug fixes.

A Hugging Face blog post describes how reordering tasks on the same GPU cluster boosted utilization by 33 percentage points. The post details scheduling optimizations that improved efficiency without adding hardware, highlighting the impact of job order on cluster performance.

OpenAI and CodeAI announced a partnership to help students build AI literacy, think critically about AI, and develop skills to use and shape it responsibly. The collaboration aims to prepare the first AI generation for a future where AI is ubiquitous.

New arXiv papers cover LLM judge instability (Wiggle Framework), privacy-preserving RAG (SEAG), multi-agent pricing, transformer expressiveness, constraint saturation, Dual-Flow Transformers, governed memory, and edge scheduling. Findings: judges flip verdicts 25-91% under pressure; SEAG hides sensitive entities with >74% accuracy; instruction following collapses beyond 5-6 constraints.

OpenAI introduced ChatGPT for Teens, designed to help teens learn and think critically with AI. It includes stronger built-in protections, healthy-use features, and additional parental controls.

OpenAI released Codex v0.148.0, adding TUI conversation export to Markdown, session forking with `codex exec fork`, draft prompts during startup, and Amazon Bedrock Runtime as a built-in provider. Bug fixes include model switching, session restoration, and sandbox restrictions failing closed on denied paths.

Hugging Face's Sentence Transformers library now supports multi-vector (late interaction) embedding models, enabling richer document representation. This update allows for more nuanced semantic search and retrieval, improving performance on complex queries.

Agentic coding tools now autonomously read repos, run tests, and open PRs, shifting from suggestion to delegation. However, a developer warns that vague instructions cause failures, and an orchestration platform analysis shows infrastructure overhead consumes 20-30% of job runtime, with pre-warming caches cutting times from 30 to 20 minutes.

Asana leveraged OpenAI's Codex to replace an outdated testing system in two weeks, a task estimated to take five years, at a cost of about $12,000. The project demonstrates the potential of AI-assisted coding to dramatically accelerate software engineering tasks.

Meta's Generative Ads Recommendation Model (GEM), the foundation model for ads recommendations on Instagram and Facebook, now trains at LLM scale on thousands of latest-generation GPUs. The company doubled end-to-end training efficiency to 20–25% Model FLOPs Utilization (MFU) while scaling training FLOPs 4x.

NVIDIA teams use ChatGPT Work to reduce manual tasks, connect fast-moving signals, and scale successful workflows globally, according to an OpenAI blog post.

openai-ruby v0.80.0 released on 2026-08-17 adds Amazon Bedrock Runtime support, plus API updates for Ultrafast tier, structured MCP, websocket errors, and separate websocket events. Bug fixes and chore improvements are also included.

OpenAI Ruby v0.79.0 released with HTTP response observability, Tapioca typing for structured outputs, WebSocket stream IDs, and Azure OpenAI v1 support. It also adds default headers, exposes request IDs, deprecates Sora video APIs, and includes bug fixes for model identifiers and timeout handling.

NVIDIA and its partners are investing in American manufacturing, supply chains, energy grids, and skilled workforces to build infrastructure for better healthcare, scientific discovery, industrial productivity, and global tech leadership.

OpenAI announced it will roll out ads in ChatGPT on Free and Go plans later this month, updating its privacy policy accordingly. Ads will initially be non-personalized, based on conversation topic and limited context. Users can opt into personalized ads later. Plus, Pro, Enterprise, Business, and Education plans remain ad-free.

OpenAI introduced GPT-Live, a system for continuous voice interaction with AI, built in six months. It uses a turnless speech model and low-latency architecture to enable faster, more natural conversations, allowing users to speak without waiting for turns.

Google and Kaggle's AI Agents Intensive, a free course, attracted 353,000 participants to build and deploy AI agents. The course aimed to teach the next frontier of AI development, highlighting the massive demand for agentic AI skills.

OpenAI released v1.1.0 of its Terraform provider, adding organization and project spend limits. The update also includes dependency bumps, performance caching for project user reads, and CI improvements. This release follows v1.0.0 and is available on GitHub.

OpenAI Ruby library v0.78.0 released, adding support for custom HTTP transports and hardening them. Bug fixes and dependency updates included. Released August 7, 2026.

The Open Secure AI Alliance, with over 120 organizations, is developing SAFE guidelines to strengthen agentic AI cybersecurity. The Linux Foundation released a Request for Comments on the Shared AI Findings Exchange, aiming to improve transparency and security as Black Hat begins.

Circles, a telco company, leverages the OpenAI API and Codex to power AI-native experiences, achieving a 22% increase in ARPU, a 9% reduction in churn, and improved development efficiency.

NVIDIA highlights that surging AI demands for massive datasets and context windows exceed system memory, requiring efficient, secure storage architectures. At the Future of Storage event, the company emphasizes the need for grounded insights from AI factories, not just more capacity.

Google announced its latest AI updates for July 2026, covering new models, tools, and research. The company highlighted advancements in natural language processing, computer vision, and responsible AI, along with updates to its AI infrastructure and developer platforms.

OpenAI released Tart 2.35.0, a virtual machine tool for macOS. The update builds Tart on macOS 26 with Xcode 27, ensuring compatibility with the latest Apple operating system and development tools. This release includes a single pull request by torarnv.

Hugging Face released speech-to-speech v0.2.11, adding a chat-completions LLM backend (OpenAI /v1/chat/completions), support for text-only and out-of-band responses, a realtime web demo, and fixes for conversation races and Windows installation. Default Qwen3 TTS switched to GGML.

NVIDIA announced Alpamayo 2 Super, an open model for robotaxis and autonomous vehicles, now available for commercial use. It addresses long-tail events by enabling situational understanding, reasoning about cause and effect, and choosing appropriate actions, going beyond object detection and motion prediction.

OpenAI publicly responded to Apple's lawsuit, calling it baseless. The company corrected claims about its employees and shared messages documenting the events. OpenAI asserts the legal action lacks merit and provides evidence to counter Apple's allegations.

OpenAI addressed recent third-party cybersecurity evaluation incidents involving its models, outlining new safeguards to strengthen AI model testing and evaluation. The company aims to improve security protocols for external assessments.

Anthropic announced that Mariano-Florentino (Tino) Cuéllar will join as Chief Global Affairs Officer. Cuéllar, a former California Supreme Court justice, will lead Anthropic's global policy and societal impact efforts, focusing on AI safety and regulation.

Microsoft Research released Orchard, an open-source framework for training and evaluating AI agents across task types. It reduces complexity and supports strong performance from smaller models by enabling researchers to reuse infrastructure, aiming to scale agentic AI research.

Anthropic published details on how its text watermarking system for Claude operates, describing the technical mechanism used to embed detectable signals in AI-generated text. The blog post outlines the approach behind identifying content produced by Claude models, aiming to increase transparency about AI-generated text provenance.

Hugging Face released a blog post titled 'State of Open Models: Summer 2026 Observations,' offering its assessment of the open-model ecosystem as of summer 2026. The post is published on the Hugging Face blog under this title, but no further details, data, or specific findings were provided in the available source content.

NVIDIA is participating in the U.S. National Science Foundation's State and Regional Artificial Intelligence Infrastructure Hubs program, launching today to expand access to advanced computing, data, software, and expertise for AI-enabled research and education. The program supports state and multistate groups, aligning with the Genesis Mission.

Google DeepMind announced Gemini 3.7 Flash, a new addition to its Gemini model family, according to a post on the DeepMind blog. The announcement page provides the model's name and introduction but does not detail specific capabilities, benchmarks, or availability in the source content provided.

OpenAI introduced new education plugins for ChatGPT Work and Codex, designed to support K–12 teachers, college educators, and students in learning, teaching, researching, and building.

Hugging Face published a blog detailing how developers can use Strands Agents, LeRobot, and Hugging Face Storage Buckets together to record, train, and deploy robotics models from a single streaming data pipeline. The integration aims to simplify moving robot data from capture through training to deployment without switching platforms, using Hugging Face's storage infrastructure as the connective layer.

OpenAI released a builder's guide explaining how startups use GPT-5.6 to build AI agents. The guide covers smarter model selection strategies and new Responses API capabilities aimed at helping developers create faster, more cost-efficient applications with the model.

Hugging Face published a blog post detailing an effort to reproduce 2,200 papers from ICML, sharing findings from the large-scale reproduction project. The post outlines what the team learned in attempting to verify and replicate results across this broad set of accepted machine learning papers, though specific quantitative findings were not detailed in the available excerpt.

Indonesia's Ministry of Communication and Digital Affairs, Indosat Ooredoo Hutchison, NVIDIA, and Universitas Gadjah Mada launched the UGM Indosat NVIDIA AI Technology Center in Yogyakarta, the country's first university-based AI center, to develop local AI talent.

Google DeepMind introduced SL2T, a new sign-language-to-text AI model designed to power sign language features for Deaf and hard of hearing users. The announcement, detailed in a DeepMind blog post, describes SL2T as a breakthrough enabling real-world product features that translate sign language into text, aiming to improve accessibility and communication for signing communities.

Google announced Sheets canvas, a new AI feature for Google Sheets that turns spreadsheet data into interactive dashboards, custom study trackers, seating charts, and other visual layouts using a simple text prompt. The feature is designed to help users quickly transform raw spreadsheet data into more usable, visual formats without manual formatting work.

OpenAI unveiled Ultrafast, a new API service tier running its GPT-5.6 Sol model up to 14 times faster than standard service. Powered by Cerebras hardware, the tier delivers up to 750 output tokens per second, aiming to speed up response generation for developers using GPT-5.6 Sol through OpenAI's API.

Allen Institute for AI introduced OlmoEarth embeddings, a new feature in OlmoEarth Studio that lets users export custom embeddings derived from satellite and geospatial data. These embeddings can be used for downstream analysis tasks, enabling researchers and developers to build custom models on top of pre-processed Earth observation data without starting from raw imagery.

Liquid AI published LFM2.5-VL-3B, a 3-billion-parameter vision-language model on Hugging Face aimed at delivering faster, more capable multimodal AI for edge devices. The blog post introduces the model as an upgrade in Liquid AI's LFM (Liquid Foundation Model) lineup, targeting improved vision understanding performance while maintaining efficiency suitable for on-device or resource-constrained deployment scenarios.

NVIDIA's GeForce NOW cloud gaming service has officially released its native Linux app out of beta. The update includes new cloud optimizations for more responsive Frame Generation while streaming, and Performance members will see higher frame rates. The timing aligns with back-to-school season, benefiting Linux and Chromebook users.

Meta's engineering team detailed an early look at Scam Alert, a WhatsApp feature designed to detect and warn users about scams—including impersonation, social engineering, and AI-generated lures—while preserving end-to-end encryption. The post outlines verifiability guarantees meant to ensure the detection system works without compromising message privacy, as scam tactics continue to evolve alongside AI capabilities.

OpenAI appointed Dali Rajic as its new Chief Revenue Officer, tasked with leading the company's global revenue organization. The role will focus on helping businesses adopt and realize value from OpenAI's AI products, according to an announcement on OpenAI's official blog.

NVIDIA announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish independent financing platforms aimed at mobilizing over $500 billion in third-party capital for AI infrastructure buildout over time.

Google's research medical AI system, AMIE, demonstrated real-time clinical video consultation capabilities in a first-of-its-kind study. The system was tested in simulated settings, showing potential for AI-assisted telehealth.

Microsoft Research introduced CARE-X, a framework for radiology vision-language models (VLMs) that combines auxiliary supervision, reward-aligned learning, and tool-augmented measurement to improve chest X-ray interpretation. It aims to move beyond report generation toward flexible reasoning and calibrated predictions.

NVIDIA founder and CEO Jensen Huang ranked No. 1 on Glassdoor's 2026 Best CEOs list, with 99% employee approval. The ranking is based on employee feedback, recognizing leadership directly from those who know it best.

NVIDIA introduced Magpie TTS, an open-weights text-to-speech model designed for low-latency multilingual voice agents. It offers full deployment control, enabling developers to build responsive voice applications across multiple languages. The release includes model weights and deployment tools, aiming to reduce latency in interactive voice systems.

OpenAI research reveals how enterprises are adopting agentic AI, using ChatGPT and Codex, and how frontier firms are pulling ahead in AI adoption.

NVIDIA announced a new 800-volt DC power architecture for AI factories, addressing the bottleneck of power delivery from grid to GPU. The design improves efficiency and scalability for next-generation accelerated computing, which demands higher rack density and more power.

IBM Research introduces SLDD (Self-Learning Data Decomposition), a method that reduces token usage for ACE (Agentic Context Engineering) by decomposing data. It achieves comparable performance with fewer tokens, improving efficiency for agentic workflows.

OpenAI sent a letter to Texas Governor Greg Abbott outlining its commitment to responsible AI infrastructure in the state, supporting reliable and transparent growth that benefits Texans.

NVIDIA is celebrating partners and open source communities advancing local AI throughout August, highlighting its latest open models, software, and tools. The initiative focuses on enabling developers to build, customize, and run increasingly capable AI agents locally, showcasing the growing ecosystem of models and applications.

Google announced new AI and agentic experiences across Google Ads and Google Analytics to simplify marketing workflows. The updates aim to help marketers evolve their strategies with more automated, intelligent tools.

RingCentral leverages OpenAI's ChatGPT Work and Codex to speed up AI product development and centralize operational intelligence across engineering and operations, according to an OpenAI blog post.

NVIDIA expands its Nemotron 3 family with Nemotron 3.5 Lightning, touted as the highest-efficiency model in its class for long-running agentic AI workloads. The release follows the earlier Nemotron models and includes NeMo Switchyard, aiming to deliver faster, smarter, and more efficient agentic AI.

A new blog post from Hugging Face discusses making knowledge distillation cost-effective for large-scale deployment. It likely presents techniques or optimizations to reduce computational overhead, enabling broader use of distillation in training smaller models from larger ones.

OpenAI introduced GPT-5.6-Cyber, a cybersecurity-specific model available through Daybreak Red for authorized vulnerability research, exploit validation, and security testing. The release aims to expand Daybreak's capabilities as the cyber defense window narrows.

Meta has released Muse Glimmer, a new AI model that is local, agentic, multimodal, and open source. The model is designed to run locally, enabling on-device AI applications. It supports multiple modalities and agentic capabilities, allowing it to perform tasks autonomously. The open-source release aims to foster community development and innovation.

Firebird, an emerging AI cloud, launched the CIS region’s largest AI factory in Armenia, powered by NVIDIA accelerated computing and Dell Technologies infrastructure. The facility establishes a new AI computing hub in the region, with NVIDIA and Dell providing the high-performance systems.

OpenAI announced it is testing ads in ChatGPT to support free access. The ads will be clearly labeled, won't influence answers, and come with strong privacy protections and user controls.

Anthropic announced improvements to Fable 5's biology safeguards, enhancing measures to prevent misuse of biological information. The update aims to strengthen safety protocols in the model's responses to biology-related queries.

Google DeepMind's WeatherNext AI model achieves a breakthrough in forecasting cyclones, improving prediction accuracy and lead times. The model leverages advanced machine learning to better anticipate cyclone paths and intensity, potentially enhancing early warning systems.

OpenAI and AWS announced that Daybreak cybersecurity models are now available through Amazon Bedrock, enabling enterprise security workflows. The integration brings OpenAI's cybersecurity capabilities to AWS customers.

Baseten is now available on Hugging Face Inference Providers, allowing developers to deploy and run models through Baseten's infrastructure directly from the Hugging Face platform. The integration simplifies access to Baseten's GPU-backed inference, offering a streamlined workflow for AI developers.

Anthropic released Claude Code v2.1.233, adding GitLab merge request URL support, an opt-in forward_user_identity setting for apps gateway, memory cgroup limits for Bash commands, and a WebFetch cache TTL config. It fixes MCP v2 stream reconnection, notification hooks, and a Windows NTLM credential-leak vector.

OpenAI CFO Sarah Friar outlines five lessons for creating an AI-native finance function, covering automated forecasting, stronger controls, and measuring AI ROI. The insights aim to guide finance leaders in integrating AI into their operations.

Mistral AI released Vibe v2.24.0, introducing an admin config layer for shared/enforced settings, a 'Default' unpinned model option, server-side default model routing, and improved subagent handoff. It also fixes drag-and-drop in terminals and speeds up /resume listing.

OpenAI announced that Model ML now uses GPT-5.6 Sol to handle finance tasks, from research and analysis to producing editable, traceable PowerPoint decks and Excel workbooks. This integration aims to increase efficiency in financial workflows.

OpenAI announced that approved Daybreak partners can now use its frontier cyber models to deliver authorized, governed cybersecurity services to customers. This move aims to put advanced AI capabilities in more trusted hands while maintaining oversight.

OpenAI is introducing premium seats for ChatGPT Business, offering higher usage limits for demanding work. Teams that sign up by August 20 receive $100 in workspace credits.

Hugging Face released speech-to-speech v0.2.12, the final 0.2.x update. It introduces Smart Turn v3.2 for better endpointing, WebRTC transport for the OpenAI Realtime API, direct audio input for audio-capable LLMs, and an optional remote-LLM proxy. The browser demo also improved with device selection and user-turn feedback.

Virgin Atlantic is using OpenAI's ChatGPT Work to accelerate research, product planning, and decision-making, helping teams connect signals across the customer journey. The airline aims to improve customer experiences by leveraging AI to analyze data and streamline operations.

Anthropic released Claude Code v2.1.232, enabling subagent forking by default, cross-session mentions via '@', and unique session names. It adds GitLab token redaction, GitLab plugin marketplace support, and fixes PowerShell and Git Bash permission bypasses. Enterprise policy validation now fails on invalid configurations.

Zapier's enterprise marketing team uses ChatGPT Work to reduce lead funnel drop-offs, build campaign assets, and automate reporting, according to an OpenAI blog post.

OpenAI released preliminary cybersecurity evaluations for its AI model Astra, outlining steps to strengthen safeguards and security controls. The announcement focuses on assessing critical cyber capabilities and enhancing protections as the model advances.

HSP GRUPPE, a tax advisory firm, is using ChatGPT Enterprise to boost productivity, improve work quality, and create more capacity for client service, according to an OpenAI case study.

Hugging Face released harbor-hf 0.1.0, an initial PyPI release of a CLI tool for planning and executing campaigns on Hugging Face Jobs, Inference Endpoints, and Sandboxes. It includes durable evidence and endpoint cleanup checks. Install via `uv tool install harbor-hf`.

NVIDIA's GeForce NOW cloud gaming service adds 26 new games in August, starting with eight this week, including World of Warships: Legends. The service is also showcasing hands-on experiences at QuakeCon in Grapevine, Texas.

Hugging Face released Transformers v5.15.0, adding support for Meta's Muse Glimmer, a 30B multimodal model for agentic use cases, plus GraniteSWA, A.X-K1/K2, and Cosmos3 Edge. Breaking changes make kernels opt-in for linear attention models and require negative values for cache cropping.

In July, NVIDIA joined over 200 companies and organizations in signing an open letter titled 'Open Weights and American AI Leadership,' arguing that AI leadership will be measured by an open ecosystem reaching every sector, not by a single frontier model.

OpenAI announced improvements to GPT-5.6 Sol in ChatGPT, citing better accuracy and consistency. The company also expanded access to GPT-5.6 Luna for free users, enabling unlimited everyday chats.

OpenAI and the American Psychological Association announced a partnership to develop evidence-based guidance, resources, and safeguards for responsible AI use, focusing on youth mental health. The collaboration aims to address the impact of AI on young people's well-being.

LangChain released langgraph-checkpoint-postgres 3.1.2, fixing a bug in delta history walking for plain-value seeds and adding conformance suite tests. The release also bumps checkpoint to 4.2.0 and includes dependency updates.

OpenAI released Signals data revealing global ChatGPT usage patterns, including country-level adoption and usage trends. The insights show how people are moving from asking questions to taking actions with the AI assistant.

LangGraph released checkpoint 4.2.0, adding an opt-in omit_expired feature to skip expired rows on read, fixing write collection in delta channel history, and updating dependencies including langsmith. The release includes multiple dependency bumps and code quality improvements.