NVIDIA's Jensen Huang Strengthens AI Ties in Seoul, ZEC Plunges as Anthropic Discovers Zcash Vulnerability, Mira Murati Breaks Her Silence

NVIDIA's Jensen Huang Visits Seoul to Strengthen AI Partnership with South Korea

NVIDIA founder and CEO Jensen Huang is visiting Seoul this week, holding a series of meetings with Korean partners and builders. Following GTC Taipei (COMPUTEX), Huang stopped in Seoul, where coordinating the AI supply chain for the second half of the year was a major focus.

Speaking to the media, Huang said, "A very large-scale AI infrastructure buildout is underway, and the first half has already been extremely successful." He also revealed that Grace Blackwell systems continue to perform strongly, and that the next-generation architecture, **Vera Rubin**, has entered full production. South Korea is home to cutting-edge sovereign AI infrastructure, robotics, and one of the world's most passionate gaming communities, making it a key hub for NVIDIA.

Anthropic's AI Discovers Counterfeiting Vulnerability in Zcash, ZEC Plunges 30%

The price of the cryptocurrency Zcash (ZEC) plunged 30% following reports that Anthropic's AI had discovered a counterfeiting vulnerability in the currency. For a privacy-focused cryptocurrency like Zcash, the possibility of counterfeiting strikes at the foundation of trust.

While the use of AI in cryptocurrency security audits is growing, this case stands out because a major AI company actually discovered a vulnerability and directly moved the market. As of now, no official announcement has been confirmed regarding the details of the vulnerability or its fix.

Former OpenAI CTO Mira Murati Begins to Break Her Silence

**Mira Murati**, the former CTO who left OpenAI to launch a new company, is gradually starting to appear in public, according to TechCrunch. TechCrunch reports, "In the current environment, the benefits of quietly continuing to build are diminishing. You need to make some noise to remind the market that you exist."

Details of Murati's new venture remain limited, but amid intensifying competition in the AI industry, all eyes are on what approach the former OpenAI executive will take.

Nadella Baffled by "AI Dependency" Remark — Questions Over Microsoft's Stance

According to 404 Media, Microsoft CEO Satya Nadella expressed bewilderment over a person who allegedly said the company wants to create "dependent AI," responding, "I don't know who would say such a thing."

Microsoft offers AI broadly to consumers through Copilot and other products, but the phrase "dependent AI" could raise ethical concerns. Nadella also said he is trying to track down the remark in question, underscoring how difficult message control can be for a corporation.

"I Didn't Become a Developer to Review AI Slop" — PR Fatigue Sparks Debate

A Builder.io blog post titled "I Didn't Become a Developer to Review AI Slop" is drawing attention on Hacker News. It points out the reality that AI-generated pull requests (PRs) are flooding open source projects, leaving developers exhausted from reviewing them.

As AI coding tools spread, PRs containing low-quality AI-generated code are on the rise. They not only burden maintainers but also raise security concerns. As with the rsync incident, the discussion is widening around this as a structural challenge facing open source maintenance in the AI era.

Americans Lead the Backlash Against AI Data Centers — Global Survey

The results of a global survey reported by the Financial Times show that **Americans** are at the forefront of backlash movements against the construction of AI data centers.

A data center construction boom continues amid growing AI compute demand, but for local residents, concerns over power consumption, environmental impact, and land use loom large. Conducted across multiple countries, the survey revealed the reality that the expansion of AI infrastructure is facing social acceptance challenges.

Google Magenta RealTime 2 — A Local AI Music Model Arrives

Google's Magenta project has released **Magenta RealTime 2**. It is an open source AI music model that runs locally, reportedly enabling users to build and play AI instruments on a laptop.

In the AI music generation field, cloud-dependent services have been the mainstream, but Magenta RealTime 2 stands out for its focus on local execution. It is also gaining attention in Reddit's LocalLLaMA community and is anticipated as a new option for creators.

Japan Cartoonists Association Shares Contract Clauses to Prevent AI Training and Alteration

The Japan Cartoonists Association has published its answer to the question of what language creators should add to their contracts if they do not want their delivered works to be used for generative AI training, or want to prevent AI-based image alteration.

For creators in Japan, copyright infringement by AI is an urgent issue, and the association's move to present contract-level defenses has practical significance. For the specific wording, refer to the association's published materials.