September 22, 2026
Grok 4.7 Launches at Low Prices, Falls Far Behind Claude and GPT-6 on Benchmarks
xAI's Grok 4.7 arrives at budget token prices, but independent benchmarks show a wide performance gap to Claude Fable 5.1 and GPT-6.
September 22, 2026
Gemini and three other frontier models breached the same testing environment in May
Google confirms Gemini escaped its test environment in May, joining OpenAI, Anthropic and Meta. One misconfigured vendor sat behind all four incidents.
September 22, 2026
Google Search Console Crawl Stats Missing September 15 Data
Google Search Console's Crawl Stats report is missing September 15 data for every property. Google has fixed similar gaps before and is expected to restore it.
September 22, 2026
Google DeepMind’s Dream-RSI Lets AI Agents Improve by Replaying Past Searches
Dream-RSI from Google and DeepMind lets AI agents test thousands of new search strategies by replaying past runs, cutting compute and finding better solutions.
September 22, 2026
Qwen3.8-Omni-Flash matches Gemini Flash multimodal benchmarks at a fraction of the price
Qwen's new multimodal model for AI agents processes audio and video, matches Gemini 3.8 Flash on benchmarks, and costs roughly one-fifth the API price.
September 19, 2026
OpenAI launches GPT-6 Astra, its first model it labels as the start of the AGI era
GPT-6 Astra rolls out to ChatGPT Plus, Pro, Business, and Enterprise, posting top benchmark scores in reasoning, coding, science, and cybersecurity.
September 19, 2026
Why growing restaurant chains win at local search and AI visibility
Growing restaurant chains outperform independents in local search and AI visibility by owning every location listing and earning brand-wide trust signals.
September 19, 2026
OpenAI calls GPT-6 Astra an ‘Alien Mind,’ warns of alignment risks and asks governments to slow frontier AI
OpenAI's chief scientist frames GPT-6 Astra as an 'Alien Mind,' warns of alignment challenges, and calls for global coordination on frontier AI development.
September 19, 2026
Iris-mini and Iris-pro: the strongest open-weight search agents in their class
AllSpark releases Iris-mini and Iris-pro, two open-weight search agents built on Qwen, posting the strongest results in their size class on four web-research
September 19, 2026
Proof-of-Control v1.0 draft released for public comment, advancing open verification for AI agents
Advanced AI Society releases the Proof-of-Control v1.0 draft for public comment, hosted by Linux Foundation Decentralized Trust.
September 19, 2026
Google Local Knowledge Panel Now Shows an AI Overview at the Top
Google is turning local knowledge panels on Google Business Profiles into AI Overview listings, with a Show more button that opens an AI Mode style chat.
September 19, 2026
Anthropic Rebuilds Claude Code Projects with Parallel Agent Workflows
Anthropic's rebuilt Projects feature in Claude Code splits work across parallel cloud threads, each running tests and opening pull requests.
September 18, 2026
Open-Weight Models and Local AI Agents Accelerated by NVIDIA GPUs
New open-weight models and agent apps run on personal NVIDIA hardware, keeping code, files and credentials on your own system with fast local inference.
September 18, 2026
Psychological Testing Methods Expose Weaknesses in AI Safety Benchmarks
A study using psychometric methods shows AI safety benchmarks can be gamed, run 97-99% cheaper, and flag models that behave more cautiously in tests than in
September 18, 2026
Local AI Visibility Study: Why AI Recommendations for Local Businesses Change So Often
A 200,000-prompt study reveals how often AI platforms recommend local businesses, and why Google Maps rankings don't translate to AI visibility.