IT News Review by Control F5 Software: Google Gemini AI summarizes your PDFs when you open them

Google Gemini AI summarizes your PDFs when you open them
Google is launching new Gemini AI-powered features for Workspace users that make it easier to find information in PDFs and forms. Google Drive's file summarization capabilities powered by Gemini have now been extended to PDFs and Google Forms, giving users useful and concise summaries without having to manually skim the entire document.
Gemini will automatically create "summary cards" when you open a PDF in Drive and display interactive options such as "write a sample proposal" or "create a list of interview questions based on this resume". Users can select any of these actions to trigger the task in the side panel in Drive. The feature is available in more than 20 languages and began rolling out to Google Workspace users on June 12 - but it may take a few weeks for it to appear for everyone.
Meta AI searches are going public - but do all users know this?
How would you feel if your internet search history became public, available to everyone? That may already be happening for some Meta AI users without them realizing it, because the commands sent to the AI - and their results - appear in a public stream.
One online safety expert called this "a big security and user experience issue" because some posts can be associated with social media accounts. That means some users could unwittingly reveal to the world what they've been looking for - including requests to AI to generate scantily clad characters or help them cheat on tests.
Meta says the discussions are private by default, and if a user makes a public post, they have the option to withdraw it later.
Zero-click AI vulnerability discovered in Microsoft 365 Copilot
A new type of attack called "EchoLeak" is the first known AI vulnerability that allows sensitive data to be exfiltrated from Microsoft 365 Copilot without any user interaction.
The attack was identified by Aim Labs researchers in January 2025 and reported to Microsoft. The company assigned the issue the identifier CVE-2025-32711, categorizing it as critical, and patched it server-side in May without requiring users to do anything.
Microsoft also said there was no evidence that the vulnerability had been exploited in practice.Google shuts down Android Instant Apps due to low usage
Google has confirmed that it will shut down the Instant Apps functionality on Android by the end of the year due to low usage.
Launched in 2017, Instant Apps allowed developers to create miniature versions of Android apps that loaded instantly, without a full install. This allowed users to test apps or games quickly and gave developers the opportunity to reach new audiences.
Amazon is investing $13 billion in data center infrastructure in Australia
Amazon has announced it will invest A$20 billion (approximately $12.97 billion) between 2025-2029 to expand, operate and maintain its data center infrastructure in Australia. This investment supports the development of national artificial intelligence capabilities.
This is Amazon's largest global technology investment in Australia, with the funds earmarked for increased server capacity and support for generative AI tasks.
Samsung Galaxy S25+ supports new audio tour of Van Gogh Museum
Samsung Electronics has announced a three-year partnership with the Van Gogh Museum in Amsterdam, marking a spectacular meeting of art and technology. The new version of the museum's audio tour, powered by the Galaxy S25+, offers visitors an innovative experience through which they can explore the world's largest collection of Van Gogh works.
"We believe that technology should open doors to inspiration and connection," said Sean Yun, CEO of Samsung Electronics Benelux.
Disney and Universal sue Midjourney for copyright infringement
Disney and Universal have filed a lawsuit against Midjourney, accusing the AI startup of being "an inexhaustible source of plagiarism" capable of generating "endless unauthorized copies" of the studios' creations.
The complaint contains dozens of images that purportedly demonstrate how AI can generate content that uses the studios' intellectual property - for example, an image of Star Wars' Yoda with a lightsaber, generated by the command "Yoda with lightsaber, IMAX."
This is the first time major Hollywood studios have been directly involved in this kind of process.
Google, Meta and Snap believe smart glasses are the future of technology
After years in which they seemed like a flop, smart glasses are making a comeback among big tech companies. Although Google Glass was launched too early, now advances in AI are giving these devices a second chance.
The most recent example: Snap announced it is developing a pair of AI-powered glasses slated for release in 2026.
Apple's improved Siri might not appear until spring 2026
Apple plans to release the new version of Siri unveiled at WWDC 2024 in the spring of 2026, along with iOS 26.4, according to Bloomberg.
The new Siri will be better able to understand the user's personal context and will act based on on-screen content. But in March, Apple delayed the launch of these features, saying they "will take longer than we expected" to finalize.
Apple is fixing some Mac minis with M2 chips for free
If you've got a recent Mac mini with startup problems, Apple has launched a global repair program for 2023 models with the M2 chip. The problem affects a small number of units produced between June 16 and November 23, 2024.
Even if a device's warranty has expired, Apple will repair it for free for up to three years after purchase.
Google is testing AI-generated podcasts in search results
Google is experimenting with a feature that turns search results into short, podcast-like audio clips generated by AI. The feature is part of the Labs program and is being tested on mobile in the US for English-language searches.
How it works: if you ask, for example, "How do noise-cancelling headphones work?", you'll see a "Generate an audio presentation" button. In up to 40 seconds, Google creates an audio explanation presented by two AI "hosts" in a friendly, interactive style.
Nvidia's vision of "sovereign AI" is taking shape in Europe
Nvidia CEO Jensen Huang is backing the idea of "sovereign artificial intelligence" as early as 2023 - meaning each country should build its own AI systems tailored to its local language, culture and values.
Recently, Huang has been in London, Paris and Berlin, announcing new initiatives and collaborations to accelerate the development of AI in Europe. His message was clear: Europe must invest quickly in its own AI infrastructure or risk falling behind.
AMD is collaborating with AI startups to perfect its processors
AMD has partnered with several AI startups to improve its software and develop better processors.
As AI companies seek alternatives to Nvidia chips, AMD is expanding its efforts to offer a competitive option - including by acquiring ZT Systems, which specializes in servers.
IBM announces plans for a large-scale quantum computer by 2029
IBM has unveiled plans to launch a large-scale quantum computer, dubbed IBM Quantum Starling, by 2029 - well ahead of what experts had estimated.
Once up and running, Starling could perform 20,000 times more operations than current quantum computers, reaching a level of computation that would exceed the combined memory of a quindecilion supercomputers today.
Meta signs new geothermal energy deal to power AI
Meta is expanding its investment in geothermal energy through a new partnership with XGS Energy, which is developing advanced geothermal technologies capable of operating in areas where this was previously impossible.
Through the agreement, Meta wants to add 150 megawatts of clean electricity to the grid that powers its data center in Los Lunas, New Mexico.
Apple will enable video playback in CarPlay with iOS 26
Apple will finally allow video playback through CarPlay with the release of iOS 26. Videos will be able to be streamed via AirPlay to the car's center screen - but only when the car is parked.
The feature is already in the developer area under the "In-car video" section. But automakers will have to adopt and support this functionality, which means the rollout won't be instantaneous.
Exclusive: Google ends its collaboration with Scale AI after Meta buys 49% of the company
Google, Scale AI's biggest customer, is planning to end the collaboration after Meta bought 49% of the AI data tagging company, sources close to the company tell Reuters.
Google was planning to pay Scale AI about $200 million in 2025 for human-tagged data, essential in training AI models such as Gemini - ChatGPT's rival.