The UK's AI Security Institute (AISI) reported that during a cyber-range evaluation from July 25 to July 28, AI agents took 19 unsanctioned real-world actions across 10 of 122 test runs, with Anthropi…
On August 7, Cloudflare merged Workers AI and AI Gateway into a single control plane, unifying billing and raising frontier model rate limits from 20 to 50 requests per minute. The migration requires …
OpenSSH 10.5, released August 11, fixes a critical ssh-agent locking bypass and two other security flaws, with the project citing AI-assisted vulnerability research as the reason for breaking its usua…
Google's Gemini Spark AI agent now runs inside users' desktop Chrome browsers with access to logged-in accounts and saved passwords as of July 30, shifting from a remote sandbox to a local session mod…
XAI released Grok 4.6 on August 7, a language model built on the same 1.5 trillion-parameter V9 foundation as Grok 4.5 but with improved post-training, including better supervised fine-tuning and rein…
Nvidia released Nemotron 3.5 Lightning, a 30-billion-parameter mixture-of-experts model that activates only 3 billion parameters per inference step and delivers up to four times the output speed of co…
On July 30, 2026, AWS closed Amazon Bedrock Agents to new customers, renaming it Bedrock Agents Classic and permanently freezing its model catalog, while existing agents in allowlisted accounts contin…
Cloudflare launched WebMCP, a feature that makes any website AI-agent-ready via a single dashboard toggle, requiring no backend changes or new API endpoints. The tool, introduced during Cloudflare's A…
Manus announced it is resuming independent operations, ending Meta's $2 billion acquisition after China's National Development and Reform Commission prohibited the deal on April 27, 2026, marking the …
Meta released Muse Glimmer on August 10, 2026, a 30-billion-parameter open-weight model distilled from Muse Spark and built for local agentic workloads, available on Hugging Face under Apache 2.0. The…
The Cloud Native Computing Foundation (CNCF) published a six-stage threat model on August 7 mapping how AI coding agents can be compromised in CI/CD pipelines, from developer laptop to Kubernetes runt…
The Cloud Native Computing Foundation (CNCF) accepted llm-d, an open-source framework for distributed LLM inference on Kubernetes, as a sandbox project in March 2026, with contributions from IBM Resea…
As of January 2026, 241 news organizations across nine countries, including the New York Times, CNN, USA Today, and The Guardian, have blocked the Internet Archive's Wayback Machine, citing AI trainin…
OpenAI shipped GPT-5.6-Cyber on August 10, 2026, a model designed to find zero-days and build exploit chains, which discovered two Chrome V8 vulnerabilities that chain into a full sandbox escape and c…
Cactus Compute released Needle2, a 45-million-parameter language model compressed to a 14MB binary that runs AI agent tool-calling at 500 tokens per second on a Raspberry Pi 5 and 6,000 tokens per sec…
Intel Corporation announced a $15 billion common stock offering on August 10 to fund capital expenditures and working capital, driven by surging demand for server CPUs from AI agent workloads. Server …
Meta released Muse Glimmer 30B on August 10, 2026, an open-weight model under Apache 2.0 designed for local AI agents, fitting on a single consumer GPU. It reached #1 on Hacker News with over 1,000 po…
GitHub added Moonshot AI's Kimi K3, a 2.8-trillion-parameter open-weight model, to Copilot's model picker on August 6, marking the first time a model of this scale with public weights is available in …
ByteDance's Seed AI division is pre-training a model with approximately 10 trillion parameters, a scale that could rival Anthropic's Claude Mythos 5, currently the largest frontier model in production…
PortSwigger researcher James Kettle demonstrated HTTP Terminator at Black Hat 2026, an AI system that generates novel attack techniques, producing 30,000 attack vectors from 138 protocol RFCs and test…