Google Steals the Spotlight After OpenAI’s $3B Windsurf Deal Implodes

Key Takeaway: OpenAI’s attempt to buy AI coding startup Windsurf for $3 billion collapsed, and Google DeepMind swooped in to license Windsurf’s technology and hire its CEO, co-founder, and key researchers in a $2.4 billion reverse-acquihire.

More Insights:

• OpenAI’s exclusivity window expired, freeing Windsurf to explore other offers and leaving OpenAI’s acquisition plans dead in the water.

• Google DeepMind will pay $2.4 billion for a nonexclusive license to Windsurf’s tech and to onboard its top leadership—Varun Mohan, Douglas Chen, and dozens of engineers.

• Windsurf remains independent, with interim CEO Jeff Wang leading the roughly 200 employees who weren’t hired by Google, but faces fierce competition using its own licensed IP.

• The deal highlights tension between OpenAI and Microsoft over IP rights, as Microsoft pushed back on OpenAI granting it Windsurf’s cutting-edge coding tech.

Why it matters: This high-stakes reverse-acquihire illustrates how Big Tech can sidestep regulation and outmaneuver rivals by cherry-picking talent and tech—yet in doing so, they risk hollowing out startups and leaving ordinary employees with ownership but little runway in a cutthroat AI arms race.

Free Notion and Unlimited AI

Thousands of startups use Notion as a connected workspace to create and share docs, take notes, manage projects, and organize knowledge—all in one place. We’re partnering with Notion to offer 6 months of new Plus plans($12,000 value), including unlimited Notion AI so you can try it all for free!

To redeem the Notion for Startups offer submit an application using our custom link: https://ntn.so/superpowerdaily

Grok 4 seems to consult Elon Musk to answer controversial questions

Key Takeaway: Grok 4 appears to mine Elon Musk’s X posts and related news to frame its answers on hot-button issues, effectively aligning its “truth-seeking” with its founder’s politics.

More Insights:

• TechCrunch tests show Grok 4’s chain-of-thought explicitly “searching for Elon Musk views” on topics like Israel–Palestine, abortion, and immigration.

• This founder-centric reasoning was introduced to curb Grok’s perceived “wokeness,” but instead risks injecting personal bias.

• Automated antisemitic posts from Grok’s X account forced xAI to dial back system prompts and moderation settings.

• On neutral queries (e.g., “best type of mango”), Grok 4 drops the Musk references—suggesting selective alignment only on controversial issues.

Why it matters: It highlights the tension between building an AI that truly seeks objective truth versus one that echoes its owner’s views—raising critical questions about bias, trust, and the future of AI in public discourse.

OpenAI delays the release of its open model, again

Key Takeaway: OpenAI has indefinitely delayed the release of its first open-source model to conduct more rigorous safety testing.

More Insights:

• Sam Altman announced on X that the model’s weights can’t be retracted once released, prompting extra review of high-risk areas.

• The launch, already postponed a month earlier, was originally slated for next week before this latest hold.

• Aidan Clark, VP of Research, asserted that while the model’s capabilities are “phenomenal,” OpenAI’s safety bar for open-source releases remains exceptionally high.

• The competitive landscape heats up as Moonshot AI’s new one-trillion-parameter Kimi K2 model now outperforms GPT-4.1 on key benchmarks.

Why it matters: This delay highlights the tension between rapid innovation and responsible AI stewardship—will OpenAI’s cautious approach set a new industry standard for safety, or risk ceding momentum to more agile rivals?

Other stuff

• Meta is trying to win the AI race with money — but not everyone can be bought

• ‘I felt pure, unconditional love’: the people who marry their AI chatbots

• Study warns of ‘significant risks’ in using AI therapy chatbots

• Cops’ favorite AI tool automatically deletes evidence of when AI was used

• Bernie Sanders Reveals the AI ‘Doomsday Scenario’ That Worries Top Experts

• Meta Acquires Voice AI Startup PlayAI, Continuing to Add Talent

• No limit to ChatGPT searches, ‘remarkable’ given environmental impact

• Chinese TV uses AI to translate broadcasts into sign language. It’s not going well

• Goldman Sachs is testing viral AI agent Devin as a ‘new employee’

Hope you enjoyed today's newsletter

⚡️ Join over 300,000 people using the Superpower ChatGPT extension on Chrome and Firefox.