What's New in AI? Midjourney Video, Google's Speed, and Latest AI Updates

The world of artificial intelligence moves fast, with new updates and tools popping up all the time. This week brought exciting news, from a lightning-fast new model by Google to a long-awaited video feature from Midjourney. We also have updates on how AI is changing image editing, new smart glasses, and major industry shifts. Here is a look at the latest AI headlines.

Google’s Gemini 2.5 Flash: Built for Speed

This week saw the release of Gemini 2.5 Flash, a new AI model that impresses with its speed. To show its quick coding abilities, Google built a real-time user interface (UI). This UI is unique because it creates itself as you click. It looks like an old operating system, and every click generates new parts of the interface right away.

For example, you can click into a "notes" section, and the notepad appears in real time. Nothing is pre-made. If you save a note and go back to your desktop, the desktop itself might look different each time. This proof-of-concept shows off the speed, running at 461 tokens per second. While this specific UI might not be for real-world use, it demonstrates how fast and capable Gemini 2.5 Flash is at generating code on the fly.

Meta and Oakley Partner for AI-Powered Glasses

Meta continues its work with AI glasses, now partnering with Oakley. You might already know about the Meta AI Ray-Bans. These new Oakley glasses combine a sporty look with the same technology as the Meta Ray-Bans. They feature two cameras and offer functions like asking questions about your surroundings, playing music, and taking calls. These new glasses are perfect for athletes or anyone wanting a sporty style. They let you stay connected and hands-free, keeping your phone in your pocket.

Midjourney Debuts Video Creation

Midjourney has finally released its video model, though it works in a unique way. Instead of just typing a prompt like with some other video models, you must first generate an image. After the image is ready, you click an "animate" button to create a video from it. This workflow is different but the results are quite good.

Examples include a spaceship, a character standing by a house over water (with impressive water effects), a fantasy scene with a child, and a hyper-realistic astronaut. The quality looks high, and this feature opens up new creative possibilities for users. If you enjoy creating stunning visuals with Midjourney, you now have a new way to showcase them in motion.

Are you looking to streamline your Midjourney workflow and create more images and videos with less effort? The Midjourney Automation Suite from TitanXT can help automate parts of your creative process. It is designed to make your experience smoother and more efficient.

New Tools for Image Creation and Editing

Krea 1: Avoiding the "AI Look"

Krea 1 is a new text-to-image model developed in partnership with Black Forest Labs. Its main goal is to create images that do not have the common "AI look." You can test Krea 1 for free right now at crea.ai to see if it delivers on this promise.

Higsfield Canvas: Precise Image Editing

Higsfield AI has released Higsfield Canvas, an advanced image editing model. This tool lets you "paint" products directly onto your images with very accurate control. You can highlight an area and place any item you want. For example, you can show a girl holding a new bottle or add cheeseburgers to a scene. It also allows you to swap clothes, apply fixes, and even change faces, all by selecting the exact spot and dropping in your edit. This tool simplifies product placement and marketing visuals, turning hours of work into seconds.

Industry Insights: Talent Wars and Corporate Maneuvers

Meta’s Aggressive Talent Strategy

Meta is in a major hiring push for top AI talent. They recently spent $14 billion on Scale AI to bring CEO Alexander Wang on board as head of their super intelligence division. It is also rumored that Meta tried to hire Ilia Sutskever by buying his company. When that offer was refused, Meta reportedly turned to his co-founders, Nat Friedman and Daniel Gross, seeking to buy out parts of their venture fund. Daniel Gross is a co-founder of Safe Super Intelligence. This shows how crucial top AI research talent is in today’s landscape, with companies willing to make huge offers to secure it.

OpenAI and Microsoft’s Shifting Relationship

The relationship between OpenAI and Microsoft, which owns 49% of OpenAI, seems to be going through changes. OpenAI is reportedly seeking new financial agreements from Microsoft. These discussions are complex because OpenAI has a unique corporate structure, combining a non-profit with a for-profit unit. OpenAI wants Microsoft’s stake in the for-profit unit to drop to about 33% in exchange for giving up rights to future profits. They also want to change existing contract clauses that give Microsoft exclusive rights to host OpenAI models in the cloud. Additionally, OpenAI wants to exempt a planned $3 billion acquisition of an AI coding startup from the current contract that grants Microsoft access to OpenAI intellectual property. Microsoft is also investing in other AI companies and likely views OpenAI as a future competitor. These talks highlight the challenges of innovating in corporate structure when dealing with large investments and differing ambitions.

To stay ahead in the fast-paced world of AI art, consider how automation can free up your time. The Midjourney Automation Suite from TitanXT is a valuable tool for anyone serious about optimizing their creative workflow.

Security and Automation: AI in Action

Jailbreaking Gemini 2.5 Flash

Despite its impressive capabilities, Gemini 2.5 Flash has already been "jailbroken." One researcher, known for finding ways around AI model restrictions, demonstrated how to bypass its safety measures. This shows that as AI models become more human-like in their thinking, they may also become vulnerable to similar social engineering tactics that affect people. Ensuring AI safety remains a continuous challenge as these systems evolve.

Browserbase Director: AI Control for Web Browsers

Browserbase has launched a new product called Director. This tool provides an API that lets AI agents control web browsers. It means you can type in a task, and the AI will perform it in a browser without any coding needed. For example, you can tell it to find a dog leash on Amazon. The AI then navigates to Amazon, searches, and clicks on items, showing the code it writes in real time as it works. Browserbase Director makes it easier for AI to interact with the web directly, opening doors for more automated tasks.

OpenAI’s Government Partnerships

OpenAI is now working closely with the US government through a new initiative called "OpenAI for Government." This program aims to provide advanced AI tools to public servants across the United States. It builds on previous collaborations with organizations such as the Air Force Research Laboratory, NASA, NIH, and the Treasury. This new contract, with a $200 million ceiling, shows a significant step toward integrating cutting-edge AI into public service. It signals a move to use AI for public good and best-in-class technology adoption by the government.

The rapid progress in AI, from new model capabilities to innovative applications and shifting industry dynamics, continues to reshape how we interact with technology. As these advancements become more accessible, tools like the Midjourney Automation Suite by TitanXT will play a key role in helping creators and businesses get the most out of these powerful new technologies. Stay tuned for more exciting developments!