
July 19 Tech news roundup OpenAI introduces an agentic AI, Copilot Vision is coming to Windows, ChromeOS and Android are merging
How did your country report this? Share your view in the comments.
Diverging Reports Breakdown
July 19 Tech news roundup: OpenAI introduces an agentic AI, Copilot Vision is coming to Windows, ChromeOS and Android are merging
DuckDuckGo introduces a filter to hide AI images from search. It is powered by filter lists created for content blockers like uBlock Origin and uBlacklist. Microsoft is testing Copilot Vision in Windows 11. Click To Do is testing a new feature called Describe Image. Google confirms that ChromeOS and Android are being merged. The latest version of the operating system will soon add a Desktop Mode, which brings a new desktop windowing system to open, group, resize multiple app windows. Google has released a new point update for Chrome 138, to patch a security issue that has been actively exploited in the wild. The issue has been tracked under CVE-20258, related to incorrect input in ANGLE and GPU. It has been patched in Chrome 138 for Windows / Mac and Linux and will be rolling out in the coming days for everyone in the update coming out in days to everyone. The new agent for ChatGPT, an AI agent that can handle tasks, uses an Deep Operator model.
DuckDuckGo introduces a filter to hide AI images from search
DuckDuckGo has introduced a new feature for its search engine, an option to hide AI-generated images from search. It is powered by filter lists created for content blockers like uBlock Origin and uBlacklist. The option, when enabled, will hide AI-generated photos when you search for imagery using DuckDuckGo’s image search. Users can optionally toggle “Show” to view the unfiltered results temporarily.
In addition to this, users can also choose to enable the “Hide AI-Generated Images” option from DuckDuckGo’s settings to have the filter enabled permanently. Doing so will save the settings in your browser until you clear your cookies. Or, you can choose to visit a special subdomain, http://noai.duckduckgo.com/, which not only removes AI images, but also disables Duck.AI and AI-assisted summaries. This is an interesting approach to reduce the number of fake images that one might encounter on the web.
Microsoft tests Copilot Vision on Windows
Microsoft has begun testing Copilot Vision in Windows 11. It is similar to the feature that was introduced in Microsoft Edge. Users can choose to share their desktop with the AI by clicking on the glasses icon in the Copilot app. It will allow the AI to see everything on your screen, and you can interact with it by typing your query, or speaking to it with your voice. Microsoft says that Copilot Vision can be useful for tasks that a user might need assistance with, from creative projects, resume, or even while playing a game.
Users have full control over when Copilot Vision is triggered, and can disable/stop it by clicking the stop button. So, it’s not going to be monitoring your screen continuously like Windows Recall. Still, Microsoft is adding AI aggressively in many of its apps, and this could bother a lot of people. Copilot Vision is currently rolling out to users who have opted into the Windows Insider Program.
Google confirms that ChromeOS and Android are being merged
It’s finally happening. After nearly a decade of speculation, Google has confirmed that ChromeOS and Android are being merged. Sameer Samat, the President of Android Ecosystem at Google, told an interviewer that ChromeOS is coming to laptops. This isn’t terribly surprising considering what Google announced when it launched Android 16.
The latest version of the operating system will soon add a Desktop Mode, which brings a new desktop windowing system to open, group, resize multiple app windows. Users will be able to access their recently opened apps from Taskbar Overflow. This could be a significant step for the evolution of Chromebooks.
Microsoft Click To Do will analyze the contents of your screen
Click To Do is a set of AI-powered features that Microsoft debuted earlier this year. It can assist you with various tasks like extracting text from images, searching the web using the text, sending an email, etc. Click To Do is testing a new feature called Describe Image.
As the name suggests, the AI will describe the contents of an image. The action appears as an option in the Photos app’s right-click context menu. Microsoft says this feature can be helpful for obtaining an overview of charts, graphs and other visual content. Describe Image is designed to use a local language model stored on the user’s computer for analyzing the images, and the descriptions are also stored locally.
Google Chrome 138 patches yet another 0-day security issue
Google has released a new point update for Chrome 138, to patch a security issue that has been actively exploited in the wild. This happens to be the fifth 0-day exploit that has been patched in Chrome this year.
The vulnerability, which has been tracked under CVE-2025-6558, is related to incorrect validation of untrusted input in ANGLE and GPU. It has a high severity rating. The issue in question has been patched in Chrome 138.0.7204.157 / 138.0.7204.158 for Windows / Mac and in 138.0.7204.157 for Linux. Google says that the update will be rolling out in the coming days to everyone.
This isn’t the only security issue that has been fixed in the update, there were 5 others, including 2 high security vulnerabilities, though Google has not revealed the details of these other flaws.
OpenAI’s ChatGPT agent can handle tasks for you
OpenAI is rolling ChatGPT agent, an agentic AI that can handle tasks for you. The new model uses Deep Research, Operator, ChatGPT to fully perform tasks from start to finish based on the user’s request. It uses its own virtual computer, and shifts between reasoning and action. ChatGPT agent will ask the user’s permission before taking some actions. It is also capable to connect to third-party services like Gmail, GitHub to assist users according to their needs.
ChatGPT Agent isn’t available for free users. It is rolling out to ChatGPT Pro subscribers, and will be rolling out to Plus and Team subscribers over the coming days. Pro users can access 400 messages per month, while other tiers get 40 messages every month. It is currently not available in Switzerland and the European Economic Area, OpenAI is working on bringing it to these regions.
OpenAI isn’t the only one that has launched an agentic AI, Google Chrome is getting one, Opera Neon browser already has an agentic AI, and Microsoft is bringing AI agents to Windows.
Xbox PC app now lets you stream your Xbox console games
Microsoft is testing a new feature in the Xbox PC app. Insiders who have opted in to test the latest builds, can now stream Xbox console games that they own, and play them on computers via the Xbox PC app for Windows 11.
Xbox Cloud Gaming (Beta) supports over 250 games including big hitters like Baldur’s Gate 3, Cyberpunk 2077, Civilization VII. The feature will be useful for people who don’t have a gaming PC to play games from their Xbox library, including console only games. But there is a catch, it’s not free. Users will need an Xbox Game Pass Ultimate subscription to access the game streaming feature. This could be a move that might add more utility to Xbox handhelds.
Xbox Cloud Gaming is available in 28 countries.