OpenClaw Finally Comes to Android and iOS

According to TechCrunch, OpenClaw (the open-source version of Claude Code) now runs on Android and iOS. The free, open-source agent program is finally making its way onto smartphones, meaning AI agents can now be used from mobile devices as well as desktop environments.

Previously available only as a CLI, OpenClaw could now reach users beyond developers thanks to the arrival of a mobile app. As one report put it, "the agent program is finally coming to your phone" — a step that could make AI agents on mobile mainstream.

Godot Engine Bans AI-Generated Code Contributions

The community behind Godot, the open-source game engine, has announced that it will no longer accept any contributions of AI-generated code.

According to PC Gamer, the Godot team explained that they "cannot believe that contributors who lean heavily on AI understand their code well enough to fix it." With the quality and maintainability of AI-generated code becoming a challenge for open-source projects, this stands out as a case where a major OSS project has drawn a clear line.

Even as AI coding tools continue to spread, the move is putting community trust and governance back in the spotlight.

A Japan-Made AI Developer Backed by SoftBank and NEC Emerges

According to ITmedia, the identity of a "homegrown AI model developer" funded by major Japanese companies including SoftBank and NEC has come to light.

Although details about the company had not been made public before, the fact that several major corporations are investing together is seen as reflecting a desire to reduce dependence on foreign AI models and to develop in-house models better suited to Japan's data and regulations. Whether this marks an important step toward building a domestic AI ecosystem remains to be seen.

Five Japanese Companies on the State of "Token Management"

Interviews with AI and HR leaders at five companies — LayerX, Rakus, Sansan, freee, and Mercari — have shed light on the reality of "token management": managing generative AI usage costs (token costs) alongside personnel expenses.

According to the ITmedia article, a common theme across the companies was the view that "AI token costs should be treated within the same framework as personnel expenses, not as conventional IT spending." As generative AI moves closer to the heart of their businesses, Japanese tech companies have begun rebuilding their cost-management frameworks themselves.

The Turn Toward "Simulated Worlds" to Improve AI Chatbots

According to Science, as gains in chatbot performance become harder to achieve, AI research is shifting toward building "simulated worlds."

With the limits of conventional text-centric training becoming visible, approaches that teach AI an "understanding of the world" through virtual environments are attracting attention. This points to a new direction in which language models learn not just to generate text, but also the rules and causal relationships of the physical world.

Backlash Mounts Over the Trump Administration's AI Redesign of .gov Sites

According to Ars Technica, the Trump administration is pushing a redesign of all .gov websites, and serious problems have surfaced with the quality of the AI-generated designs.

Critics say that the wholesale adoption of AI for government site design has created major usability and accessibility problems. The case highlights the limits and risks of AI use in public-sector digital infrastructure.

---