Home / News
AI & TECH NEWS
Full, easy-to-read articles on what's happening in AI right now — model launches, safety news, and industry moves — written in plain language, right here on ZetTool.
Roundup last updated: 9 September 2026. Click any story to read the full article.
OpenAI's newest flagship model, GPT-6 Astra, rolled out this week with strong scores on reasoning and coding benchmarks. It's also the first OpenAI model to hit the company's "Critical" cybersecurity capability level, meaning extra safety measures apply before wider release.
Read full article →Anthropic shipped updated versions of its most capable model family, aimed at coding and complex knowledge work. Fable 5.1 is the publicly available version; Mythos 5.1 remains limited to vetted organizations for security research.
Read full article →A new free learning hub from Anthropic offers courses, tutorials and badges to help everyday users get more comfortable and effective at using AI — aimed at building practical skills rather than just showcasing the model.
Read full article →Google's newest cybersecurity-focused model is being made available early to trusted defenders — governments, hospitals, and telecom providers — through a new initiative called the Fairwind Program, working with over 650 partner organizations.
Read full article →Chinese AI lab DeepSeek opened a limited-time test endpoint for V4.1 Flash, claiming it outperforms its own V4 Pro model. The test access window is short, which is typical for DeepSeek's staged rollout approach.
Read full article →Anthropic's coding assistant tool got a broad update this week, including better diagnostics, higher output limits for long tasks, and smoother integration with VS Code — along with a range of bug fixes for everyday reliability.
Read full article →Anthropic joined over 190 other AI companies in signing the EU AI Act's Code of Practice on transparency. As a result, newer AI models now add an invisible digital watermark to their text output, making it possible to check if a piece of writing involved AI — though only with special detection tools.
Read full article →A McKinsey report found that nearly a third of surveyed companies chose not to buy a software product because they could build the same feature themselves using AI coding tools — a trend that's starting to reshape how businesses budget for software.
Read full article →Design agency Wavespace released "Beyond the Chatbox," a framework arguing AI tools should move past plain chat windows toward interfaces that show their reasoning, ask for approval before big actions, and use task-specific layouts like forms and tables.
Read full article →Google pushed out its Gemini 3.8 Flash model this week, continuing its pattern of frequent, incremental model updates across the Gemini family for both consumer and developer use.
Read full article →Anthropic is working with Visa on tools that let businesses build AI shopping assistants powered by Claude, combining Visa's payment network with Claude's ability to understand and act on customer requests — aimed at merchants who want more control over how AI represents their brand.
Read full article →In a notable moment of industry coordination, all three major AI labs announced cybersecurity-focused models and early-access programs for defenders in the same week — a sign of how seriously the industry is treating AI-assisted cyberattacks.
Read full article →ZetTool's calculators and converters are built using the same everyday AI-assisted workflows this news covers. We think a short, jargon-free news roundup is a useful companion to the tools themselves — and we always link back to the original reporting rather than just repeating it.