Samsung Electronics Rolls Out ChatGPT Enterprise & Codex to Employees Worldwide
On June 21, OpenAI announced that **Samsung Electronics has deployed ChatGPT Enterprise and Codex to its employees worldwide**. It ranks among the largest enterprise deployments in OpenAI's history.
Samsung is a manufacturing giant with operations around the globe, and by combining ChatGPT Enterprise's secure environment with Codex's code generation capabilities, it aims to streamline operations and transform its development processes. As a case of organization-wide AI adoption in manufacturing, the move is likely to influence other major manufacturers as well.
China's Z.ai Open-Sources a Frontier Coding Model — the Same Day the US Bans Its Rival
Two geopolitically contrasting developments occurred on the same day.
**Z.ai**, a Chinese AI company, has released a frontier-class coding-focused model as open source. Having a high-performance coding model freely available is big news for the developer community.
On the same day, however, **Washington moved to ban a competing model**. The sight of influence-expanding open-sourcing and containment-through-export-controls advancing simultaneously symbolizes a new phase of the US-China rivalry in AI.
On Hacker News, some pointed out that "open-source strategy is becoming a geopolitical weapon." One could say we have entered an era in which the technical choice between open and closed ties directly into national strategy.
'CHIPS Security Act' Requiring AI Chip Location Tracking Gains Industry Support
In the US, **the 'CHIPS Security Act', a bill that would mandate location tracking of AI chips**, is winning support from the semiconductor industry.
The bill would create a mechanism to monitor where AI chips are being used, aiming to prevent military diversion and illicit flows to embargoed countries. This should translate into additional costs and burdens for chipmakers, yet what stands out is that the industry as a whole is actually voicing support.
Behind this lies the industry's preference for clear rules over regulatory uncertainty, along with the fact that chip tracking technology itself is becoming practical.
Power Plant Construction Races Ahead, Public Review Falls Short Behind the AI Boom — Reuters
According to an investigative report by Reuters, **power plant construction for AI data centers is proceeding at a rapid pace**, but public oversight and review have become insufficient along the way.
AI's compute demand is driving a surge in electricity consumption and placing new loads on the power grid. However, as plant approval processes are simplified and accelerated, environmental impact assessments and input from local residents are being pushed aside — a reality the report brought into sharp relief.
The question of how to balance the benefits of AI against environmental and energy policy may no longer be keeping pace with the speed of technological progress.
Gartner: '73% of Enterprises Fail to Manage Shadow AI,' Recommends a Division-of-Labor Model
A Gartner survey has revealed that **73% of domestic enterprises are failing to manage shadow AI (unsanctioned AI use)**.
With the explosive spread of generative AI, cases are growing where AI tools personally subscribed to by employees, local LLMs, and the like are used outside organizational oversight. Enterprises now face a 'hard problem' in which AI-specific risks have been layered on top of traditional shadow IT (unapproved SaaS and similar).
Gartner recommends **shifting to a 'division-of-labor model'**: instead of the IT department controlling everything, governance is designed in collaboration with business units. The firm notes that a blanket 'ban all AI' policy is unrealistic, and that practical rule-making that involves business units is essential.
Anthropic Begins Requiring Identity Verification, Igniting Debate Over AI Export Controls
Anthropic has started requiring some users to **verify their identity with a government-issued ID and a selfie**, and the move is making waves.
The story topped Hacker News that day with 534 points, drawing debate between privacy concerns and safety arguments. The requirement behind the identity checks is said to be **AI export controls**. For companies offering frontier models, knowing exactly who is using their services is becoming an unavoidable requirement in the regulatory environment ahead.
This issue also connects to the supply risk of depending on frontier models. Over-reliance on a particular company's model means that API policy changes or newly introduced identity verification requirements could directly hit developers' businesses.
---