Anthropic Opens Seoul Office, Expanding Partnerships with Korea's AI Ecosystem

On June 17, Anthropic opened an office in Seoul, South Korea, and announced new partnerships with Korean companies, startups, and research institutions. Choi Ki-young, Anthropic's Korea head, said that "innovation and safety are two sides of the same coin," signaling the company's commitment to supporting the Korean market over the long term. Companies such as WRTN and Law&Company are already putting Claude to work, and adoption is expected to expand across a wide range of industries, including finance, legal, and customer support.

Google's Medical AI "AMIE" Matches Primary Care Physicians on Complex Condition Management

A study published in Nature found that Google's conversational medical AI system, AMIE, performed on par with primary care physicians on complex condition management tasks. Through dialogue with patients, AMIE takes medical histories, forms differential diagnoses, and proposes treatment plans, and in evaluations it matched or exceeded clinician performance. That said, these are research-stage results, and further validation is required before the system can be used in real clinical settings.

White House Asks Anthropic to Fully Block Jailbreaks; Experts Say It's "Impossible"

The Trump administration is pressing Anthropic to completely prevent guardrail bypasses ahead of the re-release of its model, Fable 5. Security experts counter that it is simply not technically feasible to block every jailbreak. The dispute has once again thrown into sharp relief the fundamental question of how far AI model safety can be guaranteed through policy.

Pew Research Survey: Only 16% of Americans View AI's Impact on Society Positively

According to a new Pew Research survey, just 16% of Americans believe AI has a positive impact on society. In contrast to Wall Street's AI investment frenzy, anxiety about jobs and a lack of trust remain deeply rooted among the general public. For AI companies, figuring out how to improve public acceptance of the technology has become a critical challenge.

China to Build a Mechanism for Assessing AI's Impact on Employment

On June 17, China's State Council announced plans to build a mechanism to evaluate and track AI's impact on the labor market. Over the next five years, it will continuously survey and analyze the processes by which AI creates and replaces jobs. Concerns about AI-driven job displacement are a shared challenge across countries, and the move to establish a national-level assessment framework is worth watching closely.

Amazon Bedrock AgentCore Gains New Features to Strengthen Agent Knowledge Access and Learning

AWS has added new capabilities to Amazon Bedrock AgentCore. The update strengthens connectivity to organizational documents, the web, and licensed data, and provides production troubleshooting tools along with scalable governance features. It can be read as AWS's answer to the challenge that agent performance depends less on model intelligence than on whether agents can access the right context.

France's OVHcloud Plans a Frontier-Class LLM, Aiming to Become Europe's Second Major LLM Player

French cloud company OVHcloud has announced plans to develop its own frontier-class AI model. It aims to become Europe's second-largest LLM player — a significant move from the perspective of European AI sovereignty. While a detailed roadmap has yet to be revealed, it represents the most serious homegrown European LLM challenge since Mistral AI.

---