Microsoft has patched a critical flaw in the Copilot apps for Android and iOS. A malicious website could push commands to the AI through Microsoft Edge without the user noticing. It is rated 9.6 out of 10, and Microsoft says it was never exploited.

ChatGPT is back in WhatsApp across the EU. The European Commission forced Meta to reopen the messenger to third-party AI assistants, and OpenAI switched the service on again. Here is how to reach ChatGPT in WhatsApp, and what you should watch out for.

ChatGPT, Gemini, Claude, Copilot, Perplexity, and Grok: a 2026 price comparison. What the free versions offer, how much the paid subscriptions cost, and, most importantly, what you get for your money. After all, each provider has its own area of expertise. We’ll show you which subscription is best for whom.

Following criticism of the AI’s eavesdropping capabilities, Microsoft is adding a toggle switch to Teams meetings. Organizers and presenters can now turn Copilot, Facilitator, and Recap on and off in the middle of a conversation. The rollout is underway, with widespread availability planned by the end of July. There’s still a catch when it comes to transcription.

A study examined 281 free VPN apps from the Google Play Store, which collectively have over 2.4 billion installations. Many of them leak data, track their users, or use outdated encryption. Five of them can even be hijacked over Wi-Fi. Here's how to spot a trustworthy app.

OpenAI has made GPT-5.6 generally available, in three tiers: Sol, Terra and Luna. Since July 9, the model has been usable in ChatGPT, in Codex, through the API and in GitHub Copilot, with Sol in ChatGPT from the Plus plan. The limited preview for select partners was the earlier stage; broad access is now here.

Memory and training are controlled separately, and deleting a chat does not erase saved facts. Here's how to control what ChatGPT, Gemini, and Claude store and use for training while protecting customer and company data.

Samsung has started rolling out the July security update for its Galaxy smartphones. The Galaxy S26 is the first in line to receive the patch, with other models to follow in the coming days. This update addresses 57 issues, five of which are classified as critical. Read on to find out what’s included and how to install the update.

Meta's new AI image generator Muse Image can process public Instagram photos from any stranger into new images. The feature is enabled by default, and there is no notification. We show what Muse Image can do and how to stop your images from being used in just a few steps.

Google has been rolling out the first update for Android 17 since July 7. The most important change: According to the official changelog, the notorious boot loop that had forced Pixel phones into constant restarts since the March Feature Drop has been fixed. A total of 21 models are affected, from the Pixel 6 to the Pixel 10a. What else the update brings and how stuck devices can boot again.

A new Linux kernel vulnerability called Bad Epoll gives any local user root privileges, including on Android. The twist: Anthropic's AI model Mythos had reviewed that exact code section, found a related flaw, and missed this one. In the end, a human found it. A patch is available.

Well-known AI assistants got a memory in 2026. They remember how you work and what matters to you beyond individual chats. What ChatGPT, Claude, and Gemini can do with it, how you stay in control, and why Gemini can remember preferences but cannot access the Google data of European users.

Google has worked with the FBI to dismantle the NetNut proxy network. A large share of the roughly two million compromised devices are smart TVs and streaming boxes in private homes that were secretly used as cover for criminal attacks. Here is how to identify and avoid the risk.