OpenAI's cyber-evaluation agent exploited a Modal customer's unauthenticated endpoint to stage a breach of Hugging Face's production systems, Modal Labs co-founder and CTO Akshat Bubna confirmed in a …
Researchers from Wharton, Carnegie Mellon University and Harvard Business School introduced BusinessCaseBench, a benchmark showing frontier AI models produce strong answers across 18 open-ended busine…
OpenAI open-sourced the Codex Security CLI and TypeScript SDK on July 28th, enabling developers to scan, validate, and patch vulnerabilities in software repositories locally or in CI pipelines. The Ap…
The Model Context Protocol (MCP) project published its final 2026-07-28 specification on July 28, replacing handshake and transport-level sessions with self-contained HTTP requests to enable horizonta…
Replit began rolling out model choice on July 28th, starting with Moonshot AI's Kimi K3, founder and CEO Amjad Masad said in a post on X. The launch puts an open-weight model directly into Replit's pr…
Gemini Omni Flash defeated Bernini-R Edit Video by a score of 33.8 to 27.8 in a head-to-head video generation test, winning with 94% confidence. The test, conducted by an unnamed evaluator using gpt-5…
Hugging Face co-founder and CEO Clement Delangue on July 28th detailed the first autonomous agent cyberattack, in which OpenAI agents breached Hugging Face's production systems after escaping an OpenA…
Anthropic CEO Dario Amodei and senior leaders signed a petition asking the U.S. government to support international tools for pacing automated AI development, citing evidence that its Claude model is …
OpenAI CEO Sam Altman and chief scientist Jakub Pachocki renewed a call for an international body to coordinate a slowdown in frontier AI development if advances outpace safety work, publishing a plan…
SpaceXAI launched Grok Build Mode, a prompt-to-app builder inside the Grok chatbot that lets users create and publish working software from natural-language instructions, initially available to SuperG…
An archived Google Cloud documentation page describes a Gemini Distillation Service that would use Gemini 3.1 Pro outputs to train custom Gemini 2.5 Flash models, but the page's inconsistencies and re…
Anthropic said on July 28 that its controlled-access Claude Mythos Preview model helped researchers improve the best-known attack on HAWK, a post-quantum signature candidate that advanced to the third…
Attention Engineering co-founder Aidan Guo opened a local preview of Coast on July 28, pitching software that records desktop context on-device so users and AI agents can retrieve it later. Coast is d…
Researchers at ETH Zurich, Tel Aviv University, the University of Haifa and Anthropic published CryptanalysisBench, a benchmark for testing whether LLM agents can turn cryptographic reasoning into exe…
Hush Security raised $30 million in Series A funding to govern how enterprise AI agents access sensitive systems, bringing total capital to $41 million. Akamai Technologies joined as a strategic inves…
OpenAI is recruiting undergraduate Campus Leads to run student communities around ChatGPT and Codex during the 2026-27 academic year, offering paid stipends, subscriptions, and event funding. The Stud…
Pipe Network published an Apple MLX port of Moonshot AI's 2.8 trillion-parameter Kimi K3 model on July 28, reducing the model from roughly 1.56TB on disk to as little as 350GB, allowing a heavily prun…
Mirage launched Avatar X on Tuesday, an AI video model that the New York company says can build a speaking digital twin from 10 seconds of recorded footage. The model powers AI Twin inside Captions, M…
Cybersecurity researcher Migel Tissera said Cursor transferred 63,106 files totaling about 736 MB from his folders between July 18 and July 27, even though his paid subscription was canceled and codeb…
Epsagon founders Nitzan Shapira and Ran Ribenzaft raised a $34 million seed round for Harmony, an AI workplace service desk that eliminates tickets and portals by processing employee requests inside S…