Welcome back. The EU is building an anti-tariff alliance, reaching out to Canada and Japan after Trump threatened a 30% levy on European goods by August if they can't negotiate "better terms." Rather than immediately retaliating, Brussels is delaying countermeasures to buy time for talks while simultaneously courting other nations hit by US tariffs. Turns out threatening everyone's economy is a great way to make them all become friends.

In today’s newsletter:

• ⚕️AI for Good: AI robot performs gallbladder removal with perfect accuracy

• ⏰ OpenAI delays open-source model release, again

• 💰 The Windsurf Deal: When AI Ambitions Collide

⚕️AI for Good: AI robot performs gallbladder removal with perfect accuracy

A surgical robot trained on video demonstrations has autonomously performed gallbladder removals with 100% accuracy.

Johns Hopkins University researchers developed the Surgical Robot Transformer-Hierarchy (SRT-H), which completed eight separate procedures on realistic models without human intervention. The robot learned through imitation learning, analyzing 17 hours of surgical demonstrations and over 16,000 recorded movements, then executed all 17 complex steps required for gallbladder removal.

"This advancement moves us from robots that can execute specific surgical tasks to robots that truly understand surgical procedures," said Axel Krieger, the Johns Hopkins medical roboticist who led the research.

Unlike previous surgical robots that required pre-marked tissue and predetermined plans, SRT-H adapts to unexpected conditions in real time. During testing, researchers changed the robot's starting position and added blood-like dyes that altered tissue appearance, but the system adjusted smoothly and maintained perfect accuracy.

Key capabilities demonstrated:

• Responds to voice commands like "grab the gallbladder head" and learns from corrections

• Makes autonomous decisions about which surgical step to perform next

• Self-corrects mistakes during procedures, averaging 6.4 adjustments per operation

• Uses transformer neural networks—the same AI architecture that powers ChatGPT—to process visual and language information

The robot took about five minutes per procedure, longer than human surgeons, but demonstrated smoother, more precise movements.

Why it matters: Autonomous surgical systems could transform healthcare by providing expert-level operations in areas with surgeon shortages, reducing complications from human fatigue and delivering consistent precision. With over 5 billion people lacking access to safe, affordable surgery, AI-guided robots could democratize access to life-saving procedures worldwide.

Get an Extra Hour of Sleep with Eight Sleep’s Pod 5

If you haven’t heard of Eight Sleep, it’s time to wake up. Their smart sleep tech is changing the way people everywhere rest and recover, and their latest product may just be the most revolutionary yet: Introducing Pod 5.

Pod 5 is a high-tech smart cover that goes directly onto your mattress, instantly enhancing your existing bed with precision temperature regulation, sound control, sleep tracking, and even automatic elevation adjustment to help with snoring. Best of all, these high-tech perks have a purpose – the Pod has been clinically proven to improve quality sleep, giving you up to an hour of additional sleep every night. 

Ready to sleep smarter? Hit this link and use code DEEPVIEW to get $350 off your very Pod 5 Ultra (and don’t worry, you get 30 days to try it and no-hassle return if you don’t like it).

⏰ OpenAI delays open-source model release, again

OpenAI has pushed back the release of its highly anticipated open-source model for the second time this summer, citing safety concerns and leaving developers without a clear timeline.

CEO Sam Altman announced the indefinite delay Friday in a post on X, saying the company needs "time to run additional safety tests and review high-risk areas."

"While we trust the community will build great things with this model, once weights are out, they can't be pulled back," Altman wrote. "This is new for us and we want to get it right."

This would be OpenAI's first open-source model in years, featuring reasoning capabilities similar to the company's o-series models. Unlike proprietary models, developers could download and run it locally without restrictions.

The delays come as competition intensifies:

• Chinese startup Moonshot AI launched Kimi K2 Friday, a trillion-parameter model that reportedly outperforms GPT-4.1 on coding benchmarks

• Meta's Llama 4 models, released in April, dominate the current open-source landscape

Aidan Clark, OpenAI's VP of Research, defended the delay. "Capability wise, we think the model is phenomenal — but our bar for an open source model is high," Clark posted on X Friday.

The safety-focused approach reflects growing industry awareness about releasing powerful AI without adequate safeguards. Unlike proprietary models, open-source weights become permanently available once released.

Here’s How To Scale Your Small Biz With AI

If you’re a small business owner, the idea of maximizing efficiency, reducing costs, and growing your operation probably sounds great – the only question is, how?

Well, we’ve got two words for you: AI agents. 

Agentic AI is transforming small businesses like yours at hyperspeed, and Salesforce is here to show you exactly how with this free download. Inside, you’ll find everything you need to know about AI agents – including what high-impact areas they can assist with, how they can reduce costs, and how some of the most successful businesses are already using them to scale. 

Ready to grow? Download your free copy right here. 

• Grok’s team issues apology after AI goes off the rails

• Figma is going public

• The streaming wars come down to just Netflix vs. YouTube

• Payment processors are pushing AI porn off its biggest platforms

• SpaceX plans to pour $2B into Elon’s xAI to keep it competitive

• Xiaomi’s new AI glasses are kicking off a wave of hype in China

• Bernie Sanders shares the AI nightmare that has experts spooked

• AI-powered brain tech lets people move a robot hand with just a thought

• Chinese company Moonshot AI’s Kimi K2 outperforms GPT-4 in key benchmarks

• Accenture: Data Science Manager

• Dropbox: Senior Data Scientist, Marketing

• Finta: AI that automates your deal driven workflows from propose to close, while unlocking your networks.

• Flowpoint: AI-powered website analytics assistant for conversion insights

• Junia AI: An SEO‑focused long‑form content generator

💰 The Windsurf Deal: When AI Ambitions Collide

OpenAI's monthslong talks to acquire AI coding startup Windsurf for $3 billion collapsed on Friday over Microsoft's IP sharing requirements, paving the way for Google to hire Windsurf's CEO and key staff in a $2.4 billion licensing deal.

The failed acquisition represents a strategic defeat for OpenAI, while Google's intervention secured critical AI coding talent in one of the largest "reverse acquihires" in tech history. Windsurf's team raised concerns that Microsoft's agreement with OpenAI would force the startup to share its proprietary technology with GitHub Copilot, its biggest competitor.

Google's deal structure avoided those complications entirely. Rather than buying Windsurf outright, Google hired CEO Varun Mohan and select engineers while licensing the company's technology. The remaining company, with about $100 million left on its balance sheet, continues operating under interim CEO Jeff Wang.

Windsurf, formerly known as Codeium, had reached $100 million in annual recurring revenue within four years in the booming AI coding tools market, which is projected to grow from $4.9 billion in 2023 to $27.2 billion by 2032.

Key details of the collapsed deal:

• OpenAI first approached Windsurf in April 2025 after failing to acquire Cursor, another AI coding startup

• Microsoft's IP sharing agreement grants the company access to all OpenAI technologies, including acquisitions

• Windsurf's leadership explicitly opposed Microsoft involvement, viewing it as incompatible with their competitive strategy

• Talks collapsed when OpenAI's exclusivity period expired on Friday

The latest deal exemplifies how tech giants are using sophisticated talent acquisitions to access AI capabilities while sidestepping regulatory constraints. Google's approach mirrors similar moves by Microsoft with Inflection AI ($650 million) and its own previous licensing of Character.AI's team for $2.7 billion.

For Windsurf investors, including Greenoaks Capital Partners and Kleiner Perkins, the licensing deal provided partial liquidity while preserving equity stakes in the continuing independent company. The deal generated returns for shareholders while leaving most of the roughly 250 employees at the original company.

The coding AI battlefield has intensified dramatically, with companies like Cursor valued at $9.9 billion following its latest funding round. These tools serve as gateways to broader developer ecosystems, creating platform lock-in effects that extend far beyond their direct revenue potential.

This deal is becoming a massive stain on the industry. While the $2.4 billion headline sounds impressive, the reality is far uglier. The founders and a select group of engineers are cashing out while hundreds of employees get left behind with equity in what's likely to become a worthless shell company.

Here's the brutal math. Google now has a license to Windsurf's core technology and hired away the team that built it. The skeleton of Windsurf gets to compete against their own former founders (and top team members) plus every other player in the space. This makes Google's Character.AI deal look generous by comparison. At least there, Google had no interest in competing in AI companionship—they just wanted Noam Shazeer.

Microsoft's IP restrictions may have killed OpenAI's acquisition, but Windsurf's leadership chose this structure. They could have said no, found another buyer, or structured a deal that properly took care of their team.

As more news comes out this week, we may get a clearer picture of what actually happens to all these employees. Hopefully, they will receive some sort of compensation as well.

🤔 Your thought process:

Selected Image 1 (Left):

• “This one has real details, the little nicks and scratches that make the penguin’s small feathers not quite perfect in some areas. The flamingos are beautiful, but just slightly too perfect.”

• “The legs reflection of the flamingo did it for me”

Selected Image 2 (Right):

• “The [other image] looks like the main subjects were placed in the photo with a sharp edge around them”

• “They both looked fake to me! But since [the other image] had the penguin in an awkward pose, I thought it might be more fake than [this image].”

💭 A poll before you go

The Deep View is written by Faris Kojok, Chris Bibey and The Deep View crew. Please reply with any feedback.

Thanks for reading today’s edition of The Deep View! We’ll see you in the next one.

P.S. Enjoyed reading? Take The Deep View with you on the go! We’ve got exclusive, in-depth interviews for you on The Deep View: Conversations podcast every Tuesday morning. Subscribe here!

P.P.S. If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.