AI Leaks & Rumors 👀 - TestingCatalog (Page 11)

3 Aug 2025 · 2 min read

NotebookLM tests new customisation flow and Gmail integration

What we know so far: NotebookLM will get a new customisation flow for Audio Overviews and upcoming Flashcards and Quizzes. Gmail integration might be added to Discover options and more!

2 Aug 2025 · 2 min read

xAI tests team accounts for Grok as competition heats up

xAI is working on a new set of features which are targeting "Teams". It may also signal that Grok will get a new Team plan eventually.

2 Aug 2025 · 1 min read

Library on Perplexity will soon get support for Images and Apps

Perplexity Library is about to expand with new tabs for Images and Apps, so it will be simpler for users to access their creations.

2 Aug 2025 · 2 min read

Copilot users may soon access image Library and Copilot 3D

Microsoft is working on a new Library section for generated images as well as preparing Copilot 3D for release in Labs.

30 Jul 2025 · 2 min read

Perplexity preps platform for possible GPT-5 rollout

What we know so far: Perplexity prepares its model selector for GPT-5 release, which can hint towards a possible announcement tomorrow.

30 Jul 2025 · 2 min read

macOS app configs suggest GPT-5 Auto and Reasoning modes support

New GPT-5 leaks: Smart Mode (GPT-5) on Copilot for iOS, GPT-5-Alpha in Cursor, GPT-5-Auto and GPT-5-Reasoning in macOS app configs.

28 Jul 2025 · 2 min read

Grok will get infinite image gen and video gen with sounds

What we know so far: "Valentin" companion is getting ready for the release. xAI is also about to introduce the "Imagine" feature with image and video gen.

25 Jul 2025 · 2 min read

First look at Video Overviews in Google's NotebookLM

What we know so far: Video Overviews will appear in the form of video slides with text, images and other visuals, narrated by a female voice.

25 Jul 2025 · 2 min read

Microsoft tests turning pictures into 3D models in Copilot

What we know so far: Copilot 3D is an upcoming experiment where users will be able to generate 3D objects from any image. Currently, it appears in active development.

25 Jul 2025 · 2 min read

Microsoft tests Copilot wallet for one-click AI shopping

What we know so far: Microsoft is working on Copilot Wallet, where users will be able to add credit card and address data in order to be able to shop without leaving Copilot.

24 Jul 2025 · 2 min read

Microsoft prepares Copilot for GPT-5 with new Smart mode in development

BREAKING 🚨: The GPT-5 model is confirmed to unify underlying reasoning and non-reasoning models into one single system. Microsoft is preparing Copilot for GPT-5 release as well.

24 Jul 2025 · 1 min read

ChatGPT’s new features may turn it into a central hub for study and productivity needs

What we know so far: OpenAI is polishing the Study Together feature before the release. In addition, the macOS app may get minor upgrades to the Record Mode.

22 Jul 2025 · 2 min read

Anthropic tests Memory and MCP support for Claude mobile app

What we know so far: Claude is about to get Memory feature with a possibility to reference previous chats. Also, Artifacts Gallery and Remote MCPs support is coming on mobile.

19 Jul 2025 · 1 min read

Grok may soon get model slider with Expert, Heavy and Fast options

Grok on the web may soon get a new slider to choose between different reasoning efforts. Underlying assumptions can hint at a new Fast model, Heavy availability to SuperGrok users and the Grok desktop app.

19 Jul 2025 · 2 min read

Signs of GPT-5 emerge in Biosec repo as OpenAI nears release

OpenAI is expected to release GPT-5 soon, and the first signs of this model being available for alpha testers are emerging.

18 Jul 2025 · 2 min read

Anthropic develops interactive Study Projects to turn Claude into a tutor

Anthropic joins OpenAI and Google in the efforts to develop a study-focused feature called Study Projects. These projects can guide users through the learning process and visualise complex concepts.

17 Jul 2025 · 2 min read

OpenAI unveils Agent Mode in ChatGPT for complex task handling

Good things - Agent Mode will work as a mix between Deep Research and Operator with access to connectors, web browser and terminal. Users will be able to schedule Agent Mode tasks; Agent Mode won't be available in the EU initially.

17 Jul 2025 · 1 min read

Google tests Guided Learning in Gemini as AI education race heats up

What we know so far - Google is working on a new Gemini mode called Guided Learning. It appeared in the works ruffly at the same time as the upcoming Study Together feature for ChatGPT was leaked.