Google Quietly Buys Code from Play Store Developers for AI Training
According to a report by 404 Media, Google has been quietly purchasing app source code from Play Store developers and using it as training data for its AI models. In some cases, developers were reportedly not given a full explanation of what the code would be used for, raising questions about transparency around data usage.
The story has also drawn comments on Hacker News, where developers expressed concern about having no visibility into how their own work is handled. While AI companies' training data collection practices have long sparked copyright and ethical debates, a platform operator buying up code directly through its own store raises an entirely new set of questions.
1 in 5 Young People Consult AI Chatbots for Mental Health — RAND Survey
According to a survey by the U.S. RAND Corporation, one in five young Americans (late teens through twenties) reported using AI chatbots for mental health purposes. Usage rates are even higher among those experiencing severe psychological distress.
Separately, the "Mind Health Report" released by AXA around the same time found that more than 60% of people are seeking psychological support from AI. The story picked up 14 points and 7 comments on Hacker News, where discussion centered on AI's role in regions without access to professionals and during after-hours times, and how to balance it with support from medical professionals.
While AI-based mental health support holds great promise in terms of accessibility, concerns have also been raised about the risks of harmful advice and excessive reliance on AI as a substitute for proper treatment.
AI Outperforms Stanford Law School Faculty — Ripples in Legal Education
A study released by Stanford Law School reported that AI achieved scores surpassing those of the law school's faculty. The story drew 33 points and 18 comments on Hacker News, reflecting growing interest in AI's capabilities in the legal field.
Analyzing legal documents and searching case law have long been regarded as promising applications of AI, but these findings could influence the future of legal education itself. That said, what AI excels at is organizing and reasoning over existing legal knowledge; opinions remain divided on its ability to construct novel legal arguments or make judgments that take social context into account.
The Dangerous Reality of Shadow AI — Okta Survey Finds 95% of Executives "Trust Their Employees"
An Okta survey has shed light on the dangerous reality of shadow AI — AI tools used without company authorization. While 95% of executives are confident that "employees use AI responsibly," in reality more than half of employees use shadow AI.
More troubling still, a notable portion of shadow AI users admit entering confidential information such as login credentials into AI tools. According to ITmedia's coverage, the survey revealed a significant gap between corporate AI governance and the reality on the ground.
As AI tools continue to spread, companies need to grasp the reality of unofficial usage — alongside formal adoption support — and address the risk of information leaks.
Microsoft Build: Agent Platform Vision and Discovery GA
Alongside Microsoft Build 2026, the company announced a comprehensive vision for an enterprise agent platform. In an Azure AI Blog post by Jay Parikh, Microsoft outlined a plan to go beyond simple chatbots and realize long-running agent teams that span software development, support, finance, HR, and other business functions.
In addition, Microsoft Discovery, a platform for building and governing agentic AI workflows, has reached general availability. The platform provides a foundation for managing and monitoring AI agents across an entire organization.
Taken together with "Scout" and "Project Solara," covered in our previous article, Microsoft is building a comprehensive strategy that spans the development, operation, and governance of agentic AI. The company's key message is that "it is not the AI model alone, but the systems that run it, that truly transform businesses" — a point that carries real weight as enterprise AI adoption shifts from the PoC (proof of concept) stage to full-scale production.