Midjourney makes its video debut

Good morning, AI enthusiasts. Midjourney’s first video model is finally here, bringing images to life with the company’s signature aesthetic.

With V1 serving as a stepping stone toward a vision of an open-world simulator, the long-awaited release feels less like a catch-up and more like the start of a whole new creative direction.

In today’s AI rundown:

• Midjourney drops long-awaited video model

• AI watchdogs detail OpenAI concerns

• Automate browser tasks with natural language

• MIT study: ChatGPT’s detrimental cognitive impacts

• 4 new AI tools & 4 job opportunities

LATEST DEVELOPMENTS

MIDJOURNEY

🎥 Midjourney drops long-awaited video model

The Rundown: Midjourney just launched the company’s first video generation model, a web-only system enabling users to animate any image into 5-second clips — coming just days after Disney and Universal sued the company for copyright theft.

The details:

• V1 transforms images through either automatic animation or manual prompts, where users can describe specific camera movements and actions.

• Each job creates four 5-second clips extendable to 20 seconds, priced at 8x image costs — which Midjourney says is 25x cheaper than rivals.

• V1 can handle images from both Midjourney and external options, with video outputs having the signature feel found in the startup’s image models.

• CEO David Holz said V1 is a stepping stone towards real-time open-world simulations, which require the building blocks of image, video, and 3D models.

Why it matters: While other video models have converged on similar styles and aesthetics, V1 outputs have a vibe that holds true to MJ’s popular image models. Being I2V only and having no audio capabilities like Veo 3, V1 won’t compare directly to top rivals — but is definitely an interesting start to the company’s future holodeck vision.

TOGETHER WITH LOVART

🧠 Create like a pro with Lovart

The Rundown: Lovart is the world’s first Design Agent, redefining creativity by guiding you through the entire visual workflow — from concept to professional-grade deliverables across images, videos, and 3D assets.

Lovart's design superpowers include:

• Access to top-tier AI tools like GPT image-1, Flux Pro, and Tripo AI in a unified platform

• Use natural language to assign tasks, edit canvases, and fine-tune layers with precision

• Generate batches of designs, storyboards, or visual kits from a single command

• Tailored for all users from marketing teams to freelancers and SMBs

Try Lovart for free and transform your ideas into reality.

THE OPENAI FILES

🔎 AI watchdogs detail OpenAI concerns

The Rundown: Two AI watchdogs introduced the OpenAI Files, a hub of documents, testimonies, and information identifying potential conflicts of interest, mapping the organizational structure, and bringing transparency to governance failures at OpenAI.

The details:

• The Midas Project and the Tech Oversight Project created the collection, archiving and providing analysis on public information and testimonies.

• The report details findings in four major areas: Restructuring, CEO Integrity, Transparency & Safety, and Conflicts of Interest.

• The Files also aim to map OpenAI’s convoluted business structure, raising concerns about the details surrounding the company’s transition to a PBC.

• The initiative also published a “Vision for Change,” proposing a plan for OpenAI to meet the “exceptionally high standards” AI firms must be held to.

Why it matters: There has been no shortage of documents and reporting throughout OpenAI’s evolution from small lab to the most polarizing AI company in the world, and this latest drop brings the controversies and critiques front and center. As we inch closer to AGI, transparency on those who hold the keys is more important than ever.

AI TRAINING

🤖 Automate browser tasks with natural language

The Rundown: In this tutorial, you will learn how to use Nanobrowser's free Chrome extension with Gemini AI to automate complex browser tasks through simple natural language commands.

Step-by-step:

1. Install Nanobrowser from the Chrome Web Store and configure it with your Gemini API key from Google AI Studio

2. Set up models, e.g., Gemini 2.5 Pro for planner/validator and Gemini Flash for navigator

3. Test with simple tasks: “Find the latest releases by [company] and summarize the key features”

4. Scale to complex automation such as social media management, competitor analysis, and data collection

Pro tip: Since it uses your existing browser sessions, you can automate tasks on platforms where you're already logged in, perfect for social media management and business research.

PRESENTED BY FIDDLER

📊 2025 LLM Guardrails Benchmarks Report

The Rundown: Fiddler’s 2025 Enterprise Guardrails Benchmarks Report delivers objective data comparisons to navigate tradeoffs and help you choose the right guardrails for any LLM use case.

In the report, you’ll learn:

• Detailed breakdowns of offerings from OpenAI, Amazon Bedrock, Azure, and Fiddler AI

• Value metrics covering latency, cost, and accuracy for every application size

• Security performance across jailbreak resistance, toxicity control, and faithfulness

Explore all the data and real-world guardrails scenarios now for free.

AI RESEARCH

🧠 MIT study: ChatGPT’s detrimental impact on cognition

The Rundown: A new study from MIT just found that students using ChatGPT for essay writing showed significantly weaker brain activity and memory retention compared to those writing unaided or using traditional search engines for research.

The details:

• Researchers divided 54 Boston-area students into three groups, tracking their brain activity via EEG while they wrote SAT essays over four months.

• One group utilized ChatGPT for writing, another used Google for web search, and the third group used no resources at all.

• The ChatGPT group displayed the weakest neural connectivity and performed worse across all three categories of neural, linguistic, and scoring.

• Brain-only writers showed the strongest neural networks across creativity, memory, and processing regions throughout all sessions.

Why it matters: While a small study and spanning just one specific task, the results show the concerning tradeoffs that the convenience of powerful LLMs can bring. With AI tools quickly being integrated across the education system, this research shows how early dependence could potentially have a costly impact on developing minds.

QUICK HITS

🛠️ Trending AI Tools

• 🎥 Hailuo 02 - MiniMax’s new SOTA AI video model

• 🤖 Gemini 2.5 Flash-Lite - Google’s fastest, most cost-efficient model

• 🎨 Adobe Firefly - Creative platform now available via iOS and Android apps

• 🎆 Krea 1 - Krea’s first image model, now freely available in public beta

💼 AI Job Opportunities

• 🎨 The Rundown - Designer (Brand & Platform)

• 📊 Together AI - Marketing Analytics and Ops Manager

• 🛡️ Cresta - Forward Deployed Engineer

• 🎮 Figure AI - Visual Interface Owner, TeleOp

📰 Everything else in AI today

OpenAI introduced a new “OpenAI Podcast,” hosted by former OAI engineer Andrew Mayne, with CEO Sam Altman saying that GPT-5 should probably arrive “this summer.”

Sam Altman also alleged on his brother Jack Altman’s “Uncapped” podcast that Meta has offered $100M signing bonuses to try and poach OpenAI talent.

Higgsfield released Higgsfield Canvas, a new image editing model with advanced inpainting controls for adding products or quickly changing details of an output.

OpenAI’s research revealed a “misaligned persona” inside GPT-4o that can cause bad behavior, helping enable the creation of an “early warning system” during training.

Google introduced Search Live with AI Mode, allowing users to chat with a Gemini-powered voice search, receive spoken answers, and see linked sources in real-time.

YouTube CEO Neal Mohan said the platform is planning to integrate Google’s SOTA Veo 3 model into YouTube Shorts for creators to use “later this summer.”

COMMUNITY

🎥 Join our next live workshop

Join our next workshop this Friday, June 20th, at 4 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. By the end of the workshop, you’ll be able to confidently build and run your own automation agents using n8n.

RSVP here. Not a member? Join The Rundown University on a 14-day free trial.

See you soon,

Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team