AI News Flash | Google Trials "Audio Overview," Which Reads Out Search Results as Audio Summaries

Hello from the world of AI! 🚀Today's newsletter covers the following hot topics.　・Google trials "Audio Overview" to read out search results　・OpenAI wins an advanced AI project from the US Department of Defense — a contract worth up to $200 million　・Apple gives Image Playground a major update — ChatGPT integration advances AI image generation

GOOGLE

Google Trials "Audio Overview," Which Reads Out Search Results as Audio Summaries

Google has announced testing of "Audio Overview," a feature that uses AI to summarize search results and read them aloud, letting users catch the key points by ear instead of reading.It builds on Google's AI note-taking tool "NotebookLM" and runs on the next-generation "Gemini" model, letting users access information more flexibly and with less effort.Google says Audio Overview is "a handy way to get information when your hands are full or reading is difficult."

Source: Google

Available via Labs During Limited Testing, for Matching Searches

It is offered as a limited test within "Google Labs," with an audio-summary option shown automatically for queries the AI judges to be a good fit.Selecting it opens a simple audio player with play/pause, volume, and speed controls, plus links to the source material.Users can also rate each summary and send feedback on Google Labs overall.

An Evolution from NotebookLM and AI Overviews

"Audio Overview" is positioned as the next evolution of "AI Overviews," Google's AI-generated summary display.NotebookLM previously offered a podcast-style audio summary hosted by AI, based on materials users submitted. Extending this to search benefits users who absorb information better by listening than reading, and those who find text less accessible.This comes shortly after the Wall Street Journal reported that Google's AI features — AI Overviews especially — are cutting visits to news sites. So alongside praise for the technology, there is growing caution about AI's effect on the web's information ecosystem.

OPENAI

AI developer OpenAI has announced a contract with the US Department of Defense (DoD) worth up to $200 million to help build AI prototypes, applying frontier AI models to administrative support, data integration, and cyber defense.Envisioned use cases include helping service members access healthcare, integrating data across programs, and proactive cyber defense.OpenAI stresses that all use cases must follow its policies and usage guidelines.The DoD's statement describes "building AI capabilities to address problems critical to national security" and uses the word "warfighting," though its exact scope is unclear.

Source: Google

It is unclear whether "warfighting" covers weapons development or just administrative/logistical activities. OpenAI had previously banned military use of ChatGPT and its API, particularly weapons development, but it dropped the phrase "military and warfare" from its usage policy in January 2024.Behind this lies Western unease over China's fast-advancing large language models (LLMs). OpenAI backer and venture investor Marc Andreessen has warned that "Western AI and Chinese AI are already in a cold war."He voiced the same view on a podcast with Jack Altman, brother of CEO Sam Altman, underscoring that geopolitical AI competition is now a reality.

OpenAI and Microsoft Cross Paths on Government AI Work — Signs of Rivalry?

OpenAI's large DoD contract highlights a subtle rivalry between OpenAI and its major backer, Microsoft.Microsoft has long held numerous US government contracts and is known for high security standards, especially in defense. OpenAI has launched its own "OpenAI for Government" program and is deepening ties with agencies including national labs, NASA, the Air Force, the NIH, and the Treasury.Microsoft only received DoD authorization for Azure OpenAI Service to handle classified data in April 2024 — and OpenAI's direct DoD contract soon after symbolizes the tension between these partners-turned-rivals.Neither company has commented officially.

APPLE

Apple Gives Image Playground a Major Update — ChatGPT Integration

Apple's "Image Playground" launched as a simple creative tool for individuals, but opinions were initially split over its limited functionality and output accuracy.Many users reported mismatched results — a "hand" request reportedly produced six fingers. Even basic prompts like "elderly man" or "flower" sometimes failed, raising doubts about the tool's maturity.Still, many enjoyed its ease of use and speed.

Source: Apple

Expanded Range of Expression via ChatGPT — New Feature Announced at WWDC

At WWDC, Apple unveiled an upgraded Image Playground that strengthens image generation by integrating OpenAI's ChatGPT.Beyond the previous anime and icon styles, it now adds oil painting, watercolor, vector, anime, and print styles, greatly expanding creative freedom. A new "Any Style" mode also allows generation from more detailed instructions.The new style is labeled "ChatGPT style," making clear the AI processing runs through OpenAI. Apple stresses that input is sent to ChatGPT only with explicit consent, ensuring privacy and transparency.

ChatGPT Integration as Part of the Apple Intelligence Strategy

Bringing ChatGPT into Image Playground goes beyond convenience — it appears aimed at setting the tool apart as a serious generative-AI tool among free apps.This isn't Apple's first ChatGPT integration — it's already built into Siri, Notes, and Mail as a writing tool. Siri also calls on ChatGPT for questions it can't answer itself, for more flexible assistance.The updated Image Playground is expected to ship with iOS 26 this fall. Apple appears intent on establishing it not as a passing experiment but as a genuinely useful, everyday AI image tool.