PLUS: An AI outsmarts top mathematicians and a landmark voice cloning lawsuit

It's a new day AI Rockstars!

Google has decisively entered the AI talent race by hiring the core team from coding startup Windsurf. The move effectively blocks a planned acquisition by rival OpenAI, securing key talent for Google's own projects.

This isn't a typical acquisition; it's a multi-billion dollar deal for team expertise and non-exclusive technology rights. Does this signal a new strategy where giants can hamstring competitors without a full merger?

In today's Lean AI Native recap:

• Google's strategic hire to block an OpenAI acquisition
• A landmark voice cloning lawsuit moves forward
• An AI system out-solves top mathematicians
• China's Moonshot AI open-sources its powerful Kimi-K2 model

Google's $2.4B AI Talent Grab

The Report: Google swooped in to hire the core team from AI coding startup Windsurf, including CEO Varun Mohan, effectively blocking a planned acquisition by OpenAI. The move, announced by DeepMind CEO Demis Hassabis, sees Google paying a reported $2.4 billion for non-exclusive access to Windsurf's technology.

Broaden your horizons:

• This strategic hire is a major blow to OpenAI, which was reportedly in advanced talks for a full acquisition of Windsurf, blocking a key competitor in the escalating AI talent race.
• The new team will join Google DeepMind, focusing on boosting Google's Gemini project with agentic coding capabilities—tools that act as autonomous software development partners.
• Windsurf remains an independent company and, under the terms of the $2.4 billion deal, retains the right to license its popular AI coding technology to other firms.

If you remember one thing: This move shows that the battle for AI dominance is increasingly being fought over elite talent, with tech giants willing to pay premium prices to secure key teams. It also signals a potential new playbook where companies can gain top researchers and license technology without a full acquisition.

The Price of a Voice

The Report: A federal judge ruled that a major lawsuit proceed against AI voice startup Lovo. Two voice actors allege the company created and sold AI clones of their voices after hiring them under false pretenses.

Broaden your horizons:

• The artists originally filed a proposed class action lawsuit after being paid just $1,200 and $800 respectively for work they were told was for internal research only.
• While the judge dismissed federal copyright claims on their voices, the case will move forward on claims of breach of contract and deceptive business practices.
• One of the artists discovered the unauthorized use when he heard his own AI-cloned voice on a podcast discussing how AI could impact the entertainment industry.

If you remember one thing: This ruling signals that courts are willing to apply established legal frameworks like contract law to protect creators from misuse in the AI era. The outcome could establish a critical precedent for how AI companies must approach data rights and user consent.

AI Stumps The Experts

The Report: At a private meeting, an AI system reportedly out-solved 30 top mathematicians, showcasing a major leap in AI's ability to tackle problems once reserved for human experts.

Broaden your horizons:

• The attending mathematicians expressed a mix of astonishment and concern, acknowledging AI's power to accelerate discovery while questioning the future role of human researchers.
• A key part of the event was a live demonstration where the AI solved intricate equations in real-time, tackling problems that have long challenged the field.
• The meeting concluded with a call for greater collaboration between the mathematics and AI communities to ensure the technology is integrated ethically and responsibly.

If you remember one thing: This event shows AI is moving beyond assisting with tasks to potentially leading in fields of pure scientific research. The next frontier is how to best partner with these systems to accelerate human discovery in unprecedented ways.

China's New Open-Source Titan

The Report: Chinese AI firm Moonshot AI has open-sourced Kimi-K2, a massive model that shows impressive performance on coding and reasoning benchmarks, putting it in direct competition with top models from Western labs.

Broaden your horizons:

• The model uses a Mixture-of-Experts (MoE) architecture with 1 trillion total parameters, but only activates 32 billion for any given task, making it more efficient to run than its size suggests.
• Kimi-K2 especially excels at coding tasks, outperforming models like GPT-4.1 and Claude Sonnet 4 on benchmarks like LiveCodeBench, according to the company's evaluation.
• Developers can immediately start building with the model, as Moonshot AI provides OpenAI-compatible endpoints to access the API and has released the model weights on Hugging Face.

If you remember one thing: The release of Kimi-K2 shows that high-performance, frontier-level AI development is accelerating globally. This open-source contribution gives developers worldwide another powerful tool to build innovative applications.

The Shortlist

CoreWeave plans to acquire data center provider Core Scientific, a move to vertically integrate and solidify its position as a leading cloud for AI and HPC workloads.

Researchers uncovered "GPUHammer," a new Rowhammer attack variant that can degrade AI model performance on NVIDIA GPUs by inducing bit flips in DRAM.

Genspark boosted its file upload limit to 100MB, allowing users to analyze larger video, audio, and image files within its AI-powered research platform.

Police admitted to using an AI app that unintentionally altered an evidence photo, raising concerns about the use and understanding of AI tools in law enforcement.