Starbucks Korea Shuts All Stores Over AI-Involved Campaign—CEO Dismissed

Shinsegae Group, which operates Starbucks Korea, temporarily closed all of its stores in South Korea on June 16 in response to a problematic promotional campaign. The slogan, which the marketing team chose based on suggestions from an AI tool, sparked controversy—and reportedly, some store managers never even opened the email with the attached marketing materials before sign-off.

The campaign was pulled within hours, but the fallout was significant: the CEO was dismissed the same day. It is a strikingly emblematic case of AI suggestions being put into practice without critical human review. It also exposed a structure in which the assumption that "if AI generated it, it must be correct" rendered on-site review processes effectively meaningless.

60% of Consumers Resist Brands' "AI" Messaging

According to a new survey released by WordPress VIP, roughly 60% of US consumers react negatively to the word "AI" in brand messaging. Meanwhile, companies treat AI search as a key customer acquisition channel, creating a growing gap between consumer sentiment and corporate strategy.

The findings are drawing attention as evidence that marketing approaches touting the fact of "using AI" may actually erode consumer trust. Designing the boundary between what is done by AI and what is done by humans is becoming a critical brand-building challenge.

US DOJ Argues xAI's Unpermitted Gas Turbines Are a "National Security" Issue

The US Department of Justice has argued that the gas turbines xAI—led by Elon Musk—is operating without permits constitute a matter of "national, economic, and energy security," on the grounds that the Pentagon needs xAI. With xAI's data centers consuming massive amounts of electricity, the head-on collision between environmental regulation and AI infrastructure demand is coming into sharp focus.

It signals that AI compute has effectively become a strategic resource, now being debated in military and national security contexts.

Publishers Demand Payment for AI Companies' Scraping—or Face Lawsuits

A media industry group has announced a new policy demanding payment for AI companies' unauthorized scraping of content. In response to the issue of AI companies collecting content without permission and using it as training data, publishers are taking a "pay up or see you in court" stance.

AI companies have so far claimed their practices fall within "fair use," but content providers appear to be reaching the limit of their patience. How to structure compensation for AI training data is a critical issue that will shape the survival of the media industry.

NVIDIA Blackwell Tops Every Category in MLPerf Training 6.0

On June 16, NVIDIA announced that its Blackwell platform recorded the fastest training times across all categories in MLPerf Training 6.0, the AI industry's standard benchmark. It demonstrated large-scale training on NVL72 systems using 8,192 GPUs, becoming the only platform to submit results in all seven benchmarks.

In an era where the scale and efficiency of training infrastructure directly govern the pace of AI model advancement, NVIDIA's dominance has once again been borne out by the numbers.

Plaud's AI Meeting-Notes Business Reaches $100M ARR

Plaud, which sells AI meeting transcription devices, announced that it has shipped more than 2 million hardware units and that its software business has surpassed $100 million in annual recurring revenue (ARR). It is drawing attention as a case where a hardware-plus-subscription model succeeded in a market that once looked like a niche: automating meeting notes.

It shows that AI features are no longer mere software add-ons—combined with dedicated devices, they are creating markets of their own.