Today, we’re introducing Gemini 4 Neon

A South African developer is writing an operating system from scratch in C and assembly, from a UEFI bootloader and SMP kernel to a compositing desktop. Win32 is its native API, and the first release is targeted for August 8, 2028.
Key points
Stardock today released Fences 7, the latest version of its Windows 11 Desktop organization solution, with prices starting at $10.
The post Stardock Release Fences 7 appeared first on Thurrott.com.
By Pramin Pradeep, Co-founder & CEO, BotGauge AI
For most of the last few years, my job was making software prove it works. At BotGauge, we built autonomous QA that tested our customers' apps before every release. Then those same customers started shipping AI agents, and the old playbook stopped being enough.
A test suite asks one question: does this do what it should? An agent needs a second one: what will it do when someone wants it to misbehave? That second question is red teaming. It's why we launched our agent red-teaming platform on October 28, 2026 and the developer community noticed: it finished #3 Product of the Day on Product Hunt.
Red teaming isn't new. It started as military war-gaming, became a staple of security teams paid to break in, and with LLMs turned into jailbreaking chatbots.
Agents raise the stakes because they act. They call tools, read inboxes, write to databases and hand tasks to other agents. A bad chatbot answer is embarrassing. A bad agent action is an incident.
And adoption is outrunning testing. Gartner predicts 40% of enterprise applications will feature task-specific AI agents by the end of 2026, up from under 5% in 2025. Yet most teams I talk to test agents like APIs: happy paths, a few edge cases, a vibe check. None of that simulates an adversary.
OWASP ranks prompt injection as the top risk for LLM applications, but for agents it's only the front door. Here's what we see:
• Indirect prompt injection: instructions hidden in a web page, PDF, email or support ticket the agent reads mid-task.
• Tool misuse: tricking the agent into calling a legitimate tool with harmful arguments, like a refund, a delete or an outbound email.
• Data exfiltration: coaxing the agent into leaking customer data, system prompts or credentials through its outputs or tool calls.
• Goal hijacking: slow, multi-turn manipulation that drifts the agent away from its task without tripping any single-message filter.
• Multi-agent cascades: one compromised agent passing poisoned context to others that trust it by default.
Our autonomous QA product had paying customers and a working business. But functional testing was maturing fast, and the problem our customers kept raising wasn't "does it work?" It was "can I trust it in production?"
So we rebuilt BotGauge as an Agent Control Plane, and made one calls early:
● Red teaming at the core, not a checkbox. Observability tells you what happened. Red teaming tells you what could happen, before a user or attacker finds it.
The goal is simple: attack your agent the way an adversary would, every time it changes, and show you exactly where it broke.
• 200+ custom attack strategies spanning injection, tool misuse, exfiltration, goal hijacking and jailbreaks, run against your real agent, not a sandboxed copy of the model.
• Full tracing: every failed attack links to its trace, so you see which input, which step and which tool call went wrong.
• Evals that persist: results are scored and compared across releases, so a fix doesn't quietly regress next sprint.
• Fits your stack: it sits on top of runtimes like LangGraph and AWS Bedrock AgentCore instead of replacing them.
Your agents will be attacked. Regulators are already moving: NIST's AI risk guidance and the EU AI Act both call for adversarial testing. The only real question is who finds the holes first, your team or someone else. I'd rather it be you.
Pramin Pradeep is co-founder and CEO of BotGauge AI, an Agent Control Plane for testing, observing and governing AI agents in production. BotGauge is backed by Surface Ventures, IA Seed Ventures and Saka Ventures. *
Sources
This story was distributed as a release by Jon Stojan under HackerNoon’s Business Blogging Program
Welcome to HackerNoon’s Meet the Writer Interview series, where we learn a bit more about the contributors that have written some of our favorite stories.
My name is Naresh Waghela, and I am a Senior SEO Executive, technology writer, and digital marketing professional. I specialize in SEO strategy, content optimization, digital growth, and emerging technologies.
I have a strong interest in Artificial Intelligence, Large Language Models (LLMs), Generative AI, and how technology is transforming the digital landscape. I enjoy researching complex technical subjects and turning them into practical, easy-to-understand content that helps readers learn something valuable.
My latest HackerNoon story, "Making Local LLMs Usable on 8GB GPUs," explores how developers can run and optimize local Large Language Models on GPUs with limited VRAM.
The goal was to demonstrate that working with AI models locally does not always require expensive, high-end hardware. With the right optimization techniques, developers can make AI more accessible, practical, and affordable.
I write about a variety of topics, including Artificial Intelligence, SEO, digital marketing, Generative AI, technology trends, and online visibility.
I am particularly interested in the intersection of AI and search, including Answer Engine Optimization (AEO), Generative Engine Optimization (GEO), and how AI-powered search is changing the way people discover information online.
My writing process usually starts with research. I explore a topic, understand its technical aspects, review available information, and identify what readers would find most useful.
I then organize my ideas into a clear structure and focus on making the content informative, practical, and engaging. I also spend time reviewing and refining the article to ensure accuracy and readability.
One of the biggest challenges is simplifying complex technical concepts without losing their technical accuracy.
Technology evolves quickly, especially in AI and machine learning. It takes continuous research and learning to ensure that the information remains relevant, reliable, and useful to readers.
I want to continue growing as a technology writer and digital strategist while exploring the possibilities of Artificial Intelligence.
One of my long-term goals is to contribute meaningful insights to the global technology community and help bridge the gap between complex AI innovations and practical real-world applications.
I genuinely enjoy exploring new technologies, experimenting with AI tools, and discovering interesting ideas that can turn into useful projects. Sometimes, what starts as casual curiosity becomes hours of research and experimentation.
I enjoy spending quality time with my family and exploring new places whenever I get the opportunity. It gives me a refreshing break from professional responsibilities and helps me maintain a healthy perspective.
Readers can expect more content around Local LLMs, Generative AI, practical AI implementations, AI-powered search, and the future of digital marketing.
I am particularly interested in exploring how developers and businesses can use AI efficiently without always depending on expensive infrastructure.
HackerNoon is a valuable platform for independent technology writers. It provides an opportunity to share ideas, reach a global audience, and become part of a community that values technology, innovation, and knowledge sharing.
I appreciate the platform's support for writers and its efforts to make technical knowledge accessible to a wider audience.
I believe that learning should never stop, especially in the technology industry. Every new development brings an opportunity to explore, experiment, and share knowledge.
Thank you to the HackerNoon editorial team and community for providing a platform where writers can share their ideas and contribute to the global technology conversation.
Welcome to HackerNoon’s Meet the Writer Interview series, where we learn a bit more about the contributors that have written some of our favorite stories.
My first job, in 2015, was finding out what people really think of things: market research, then reputation analysis for a national lottery brand. Everyone has a view of a lottery. Nobody is neutral. Then I moved to Istanbul and co-founded an e-commerce business selling Turkish goods directly to customers. That is where I learned what I now tell founders: you can be the voice of your company, but the product needs its own accounts, its own email and its own search presence from day one.
Later I worked on a large language model project. From that side you learn two things quickly. An answer is never as fixed as it looks: while a model is still being worked on, the same question can come back noticeably different from one round to the next. And geography matters, just as it always did in search. The same question asked from two countries can return two different pictures of the same company.
That path explains where I am now. I'm a partner at BARS Agency. We build and promote digital products, work on reputation and on how people and companies are presented in AI search. Geometrika started there as a tool we built for ourselves and our clients, to see how AI assistants describe a brand and which sources they lean on, so that strategy could start from data instead of guesses. Exhibo, a global art gallery and culture platform where I'm co-founder and editor-in-chief, asks the same question about culture: which galleries, museums and places become visible online and which quietly drop out.
On paper, these look like different careers. It's one question asked in different rooms: how does something get seen and why do people believe what they see?
I studied art history and I'm studying psychology now, which covers both halves of that question. I lived in Istanbul for several years. These days I live between Doha, Dubai and Egypt.
My latest Top Story asked why a two-hour transfer in the Alps can cost more than the flight that got you there. It belongs to a series I keep coming back to: hidden economics, the prices everyone pays and nobody explains. The newest piece in it is about film rebates in the Middle East. The detail I can't stop thinking about is that a desert now comes with a percentage attached. Every line producer knows the rates by heart. Almost no government in the region publishes what it has actually paid out.
The method doesn't change from piece to piece. Start with a number that should be public. Find out whether it is. If it is, read every row of it.
On HackerNoon it's mostly search, AI and the psychology around them. Somewhere along the way the psychology pieces and the search pieces turned into the same subject. People tend to accept a coherent, confident explanation without checking how it was put together. Memory does it to us: a manager sincerely remembers backing the idea that won. AI answers do it for us: one fluent paragraph, assembled from sources of very different quality, reads as settled fact.
Elsewhere it's culture. At Exhibo I write about art, architecture and the stories attached to objects and places. One pattern keeps coming back: fiction borrows a real place and forgets to give it back. Tatooine is in Tunisia. Agrabah is stitched together from real cities.
The subjects wander. The question underneath doesn't: what is actually going on beneath something everyone takes at face value?
I start with a table. Whatever I want to know, row by row. Half of it usually fills in an afternoon. The piece lives in the cells that refuse to fill.
Then I go to the primary documents, which in this region often means a French PDF, a ministry decree or a press release that is generous with every number except the one you need. When I find a dataset I add it back up against its own totals before I trust a single figure from it.
The research eats the writing. I can spend a week finding out that a number almost never exists in public. That week is invisible in the final piece. Finding out that something isn't there is a finding too. It just takes longer to write down.
The other challenge is restraint. When you've read seven years of a ledger, you want to show the reader every row. The reader wants three.
I want companies to be able to see how AI systems describe them and where that picture comes from, the way they once learned to read their own search results. Most still check it by asking ChatGPT about themselves once and hoping for the best. And what actually moves those answers is rarely what anyone expects. Start a podcast and the assistants often pick it up. I once shared a few of my own photos of desert monasteries. Within weeks the AI version of me had become a cultural heritage contributor and was drifting towards "philanthropist". Nobody has ever complained about that particular error.
I want the same for culture. A serious gallery in Doha or a small museum in Cairo shouldn't disappear from what people find online simply because the information about it was never made easy to find.
Trips planned around exactly one thing I want to see. Lately it's usually a ruin with a desert around it, though it can just as easily be a coastline or a forest. Everything else is allowed to happen on the way, as long as there's European coffee somewhere on the route.
And when the list runs out, I go out into the desert just to look at nothing for an hour. It is the only place where my phone and I agree to ignore each other.
Languages. Spanish, Arabic and, less gracefully, Turkish. Spanish feels familiar, at least in its logic. Arabic and Turkish don't. Many Arabic words are built from three-consonant roots, so one root gives you a writer, a book and an office. In Turkish a whole sentence can be a single word thirty letters long. Both break my brain a little.
Translators and AI do fine with a menu. Face to face, they still get lost. So I'm doing it the long way.
And art, which stopped being a hobby the day it became a job. I will still talk Monet and Manet with anyone who confuses them.
First, an experiment I've wanted to run for a while: asking every major AI assistant who I am, then tracing the wrong answers back to where they came from, when they came from anywhere at all. A machine's idea of you is assembled from places you have forgotten you ever wrote.
Then more hidden economics: who pays for a free museum. The British Museum opened free in 1759, on a collection a public lottery had helped pay for. In 2026 the Louvre and the Grand Egyptian Museum both raised their prices. Someone always pays. The question is who. And whether they know it.
Where else would a story about airport transfers in the Alps sit next to one about falling in love with chatbots? I've stopped asking which of them is the tech story.
If you ever go looking for a number and it isn't there, write that down. The absence is usually the story.