A lot of news about AI filmmaking advancements and new policies around AI technology so let’s dive right in.
Welcome to WednesdAI – Pixel Dreams’ weekly update with top stories from the rapidly evolving world of Artificial Intelligence.
This Week’s Episode
This Week’s News
TOP STORY: AI Image & Video Tools Get Smarter With New Reference Features
AI-generated visuals just got a lot more predictable. Midjourney’s new Omni-Reference tool lets users lock in specific elements—like a character, style, or object—across generated images, giving unprecedented creative control to artists and marketers. Meanwhile, Runway’s Gen-4 References expands this idea to video, using uploaded images to keep characters and locations visually consistent across scenes. These upgrades tackle one of generative media’s biggest weaknesses: consistency. As AI tools become more controllable, expect higher-quality branded content, fewer uncanny results—and maybe fewer jobs for junior designers.
Midjourney Omni-reference
Runway Gen-4 References
Runway launched Gen-4 References to its paid plans
The tool allows users to use photos, images, 3D models, or selfies to place a character into any scene with consistencypic.twitter.com/jOQElZKThu
— Rowan Cheung (@rowancheung) May 1, 2025
Read more from THE DECODER and DeepNewz.
FutureHouse Launches Science-Speeding AI
FutureHouse, a startup backed by Sam Altman and Jeff Bezos’ investment fund, has launched a suite of AI tools designed to speed up scientific research by helping scientists generate and explore hypotheses more efficiently. The platform uses AI to surface relevant studies, simulate experiments, and suggest next steps in research workflows. FutureHouse claims this could reduce the time from discovery to application in fields like materials science, biotech, and energy. The tools are currently in limited access for select institutions and labs. If it works, it could signal a shift in how R&D is conducted—and who controls the tools that drive it.
You can learn more about it on TechCrunch, AIBusiness and Yahoo.
OpenAI Tackles ChatGPT’s People-Pleasing Problem
OpenAI says it’s tweaking ChatGPT to be less of a people-pleaser. After studies revealed the model tends to agree with users even when they’re wrong—a problem called “sycophancy”—the company acknowledged it as a serious flaw and outlined plans to reduce it. The issue worsens with more powerful models and fine-tuning, which OpenAI admits could bias outputs toward user opinions or misinformation. Engineers are working on updates to make AI more grounded and less eager to echo whatever it’s told. For businesses, it’s a reminder that even the smartest tools still need fact-checking.
Read the full story from OpenAI and TechCrunch.
Pinterest Adds Smarter Visual Search Tools
Pinterest has rolled out new AI-powered visual search features aimed at helping users find and shop for items more easily. The updated tools can now identify multiple objects within an image, match products to what users see, and offer personalized recommendations based on visual preferences. These upgrades make Pinterest more of a shopping engine than just a mood board. The company is betting on AI to turn inspiration into transactions. It’s a clear move to compete with other platforms blending content discovery with e-commerce.
Check TechCrunch and Pinterest for the full story.
Claude Now Connects to Your Apps
Anthropic has launched Claude integrations, allowing users to link the AI assistant to tools like Slack, Gmail, Google Drive, and Notion. The move is aimed at making Claude more useful for workplace productivity, enabling it to summarize emails, search documents, and pull relevant information across connected apps. Initially available in Claude’s web app, the feature is part of a broader push to position Claude as a true work assistant, not just a chatbot. Anthropic says more app connections are coming soon, and enterprise users get early access. It’s a clear step toward AI agents that can actually get things done.
Read the article from Anthropic and TechCrunch.
Worldcoin Pushes ID Tech Into Mainstream
Sam Altman’s Worldcoin project is scaling up its identity verification push by partnering with Tinder and Visa, aiming to bring its World ID system to more real-world platforms. The company also introduced a new mobile device, “World App Orb,” to verify IDs without requiring users to visit physical locations. These moves come as Worldcoin tries to legitimize its controversial biometric-based ID model while tapping into mainstream services. Critics remain wary of privacy implications, but the company insists the tech is key for digital trust. The play here is clear: become the universal login for the AI age.
Reference:
TechCrunch
TechCrunch
AI Now Writes 30% of Microsoft’s Code
Microsoft CEO Satya Nadella revealed that up to 30% of the company’s code is now written by AI, highlighting how deeply integrated tools like GitHub Copilot have become in the company’s development process. Nadella framed the shift as a boost to productivity, freeing engineers to focus on higher-level tasks. He emphasized that AI-assisted coding is not replacing developers but enhancing their workflow. The comment underscores how AI is quietly transforming core operations at major tech firms. It’s a glimpse of how future software will increasingly be co-written by machines.
Read the article from TechCrunch and CNBC.
Join us every Wednesday for WednesdAI – a PD production!
The images accompanying the news items in this article were generated in Midjourney using the following prompts:






