🧠 OpenAI loses more top talent to Meta

Welcome, AI enthusiasts

OpenAI just lost four more top AI researchers to OpenAI. And as Mark Zuckerberg lures away top research talent to Meta, OpenAI executives say they're 'recalibrating comp,' according to an internal memo. Let’s dive in! 

In today’s insights:

• OpenAI loses more top talent to Meta

• Google’s new AI makes DNA readable

• Velvet Sundown: The AI band with 400K Spotify listeners

Read time: 4 minutes

LATEST DEVELOPMENTS

META
🧠 OpenAI loses more top talent to Meta

Evolving AI: OpenAI loses four more top AI researchers to Meta.

Key Points:

• Meta hired at least eight senior OpenAI researchers, including leaders in AI reasoning and vision.

• OpenAI’s leadership responded with emotional internal memos and a push to improve pay and perks.

• The new hires join Meta’s push to build advanced general-purpose reasoning models.

Details:

Meta’s aggressive hiring spree has hit OpenAI hard. In recent weeks, Meta pulled in eight top researchers working on core AI projects. In a Slack memo obtained by Wired, OpenAI’s Chief Research Officer Mark Chen did not mince words describing what the situation feels like. “I feel a visceral feeling right now, as if someone has broken into our home and stolen something,” Chen wrote. The memo acknowledged the wave of high-profile exits and reassured employees that the leadership team, including CEO Sam Altman, has been working “around the clock” to stop the bleed. Chen said they were “recalibrating compensation,” and exploring “creative ways to recognize and reward top talent.”

Why It Matters:

It's starting to feel like a recurring story, but Meta has managed leading AI researchers from OpenAI on a daily basis. OpenAI is quite literally scrambling to stop the bleeding of top talent. In a now-deleted tweet, one OpenAI team member even said he was "really disappointed" that leadership couldn’t keep these important, hardcore researchers. Zuckerberg is serious about building a strong team and being ready to compete. It feels like we’re watching a live chess game for the future of AI, where every researcher switching sides is a major piece taken, and each move could decide who controls the whole game in the end.

GOOGLE
🧬 Google’s new AI makes DNA readable

Evolving AI: Google DeepMind has unveiled AlphaGenome, a new AI model that helps researchers understand how changes in our DNA affect gene activity.

Key Points:

• AlphaGenome predicts how small DNA changes impact molecular processes, like turning genes on or off.

• It builds on DeepMind’s earlier success with AlphaFold, which predicted protein structures and won a Nobel Prize.

• The tool could speed up research on diseases by replacing some lab experiments with virtual predictions.

Details:

AlphaGenome is DeepMind’s latest step in using AI to understand biology. The model can predict how altering single letters in DNA affects gene activity. Scientists usually need complex lab experiments to answer these questions, but AlphaGenome offers fast, virtual predictions instead. Researchers see it as the most powerful tool yet for modeling genetic changes and their effects. It won’t be used for personal DNA predictions but could help explain rare diseases and guide cancer research. Google plans to keep it free for noncommercial use and share full model details soon.

Why It Matters:

For years, we’ve had all this DNA data just sitting there, and no one really knew what to do with most of it. Like, we could read the letters, but, okay… then what? Now, with AI like AlphaGenome, researchers can finally start figuring out what those tiny changes in DNA actually mean, way faster than before. And honestly, this could open doors to new treatments or even help us understand rare diseases that no one could crack before. It’s kinda wild to think we might move from trial-and-error labs to basically “testing” things inside a computer first.

SPOTIFY
🎸 Velvet Sundown: The AI band with 400K Spotify listeners

Evolving AI: A new band called The Velvet Sundown is gaining attention for being AI-generated, and they already have over 400,000 monthly listeners on Spotify.

Key Points:

• The band launched less than a month ago and has two albums out.

• Their members and images show all the signs of being AI-made, raising questions online.

• Platforms like Deezer are starting to label AI music, but Spotify doesn’t require disclosure.

Details:

The Velvet Sundown describe themselves as a psych-rock group with a cinematic, dreamy sound. They list fictional band members in their bio and even include a fake Billboard quote. Fans on Reddit noticed they appeared suddenly in playlists and found almost no background info. The band’s Instagram photos also look AI-generated, which adds to the mystery. This case highlights a bigger trend: AI-made music is flooding streaming services. Deezer estimates about 10,000 AI tracks are uploaded daily. Spotify doesn’t ask creators to reveal if AI is used. Big artists like Nick Cave and Elton John have warned about the risks of AI in music. Some fear it could take income from human musicians and hurt creativity.

Listen to The Velvet Sundown on Spotify here →

Why It Matters:

AI music is blowing up and it’s making us ask real stuff, like does it matter if a song is made by a person or a bot? And who should get paid when streams go to AI songs instead of human artists? Some folks love the tech, others say it’s ruining creativity and stealing voices. We’re kinda at this big moment where music, money, and tech all clash and no one really knows how it ends.

 👀 Click on the image you think is real

QUICK HITS

🇨🇳 China’s biggest public AI drop since DeepSeek, Baidu’s open source Ernie, is about to hit the market.

🤝 ‘Big, Beautiful Bill’ AI provision brings together an unexpected group of critics.

⚽ A bumbling game of robot soccer was a breakthrough for embodied AI.

🤖 MIT CSAIL researchers combined GenAI and a physics simulation engine to refine robot designs. The result: a machine that out-jumped a robot designed by humans.

🔌 OpenAI turns to Google’s AI chips to power its products, source says.

🎥 MrBeast removes YouTube AI tool after backlash.

🌊 Merging AI and underwater photography to reveal hidden ocean worlds.

☢️ Three Mile Island nuclear plant fast-tracked to reopen for AI energy demand.

📈 Trending AI Tools

• 💬 ChatDOC – AI tool that helps you analyze and ask questions about PDF documents and research papers (link)

• 🎥 Runway Gen-4 – Video generation tool that turns text or images into realistic video clips (link)

• 🎨 Ideogram – AI design tool for creating logos, posters, and visual content quickly (link)

• 🔎 Glean – AI-powered workplace search tool that finds info across your company docs and apps (link)

• 🗂️ Mem – AI-powered workspace that organizes your notes and helps you find and connect ideas fast (link)