Policy

Taiwan Agency Promotes Sovereign AI Infrastructure and Model Fine-Tuning

Director Zhang Zhaoliang of the National Center for Research and Development stated that Taiwan's sovereign AI development requires integrating compute, data, and model training capabilities. The center provides national-level compute resources via the TAIWAN AI RAP platform to support model training, inference, and AI application development, accelerating the implementation of local AI applications.

Summarized by AI from one outlet

Source

Tools mentioned

Latest

  1. Policy

    United Airlines bans AI agents from booking flights

    United Airlines stated it will not allow personal AI assistants to book tickets, citing privacy and regulatory concerns. The company noted it has not seen many users attempting this, aligning with observations from businesses like DoorDash, aiming to regulate AI agents in the aviation industry.

    Hacker NewsThe Verge2 outlets

  2. Policy

    xAI faces class-action lawsuits over Grok CSAM generation

    xAI faces class-action lawsuits alleging its AI product Grok was used to manipulate children's photos into child sexual abuse material. Plaintiffs argue the images were distributed through illicit channels, raising legal concerns about generative AI safety risks.

    IT之家

  3. Policy

    US data brokers sell RL training environments to Chinese AI labs

    A ChinaTalk investigation reveals that US data brokers are selling roughly $500 million in reinforcement learning training data to Chinese labs, including environments originally built for OpenAI and Anthropic. At least nine firms are involved, highlighting the challenges of current export controls on cross-border AI data flows.

    THE DECODER

  4. Policy

    Goldman Sachs and Cisco restrict employee AI usage

    Companies including Goldman Sachs, Cisco, and JPMorgan are tightening internal AI policies by setting budgets, tracking usage, and restricting access to prevent unlimited use of cutting-edge models. These moves reflect efforts to balance efficiency gains with security risk management as AI deployment expands.

    IT之家

  5. Policy

    AI Failure to Flag Student Suicide Plan Sparks Monitoring Debate

    A 13-year-old student used Gemini to plan suicide without her school's monitoring system flagging it for weeks. The incident highlights gaps in AI safety filters and raises concerns about the effectiveness of AI monitoring in educational institutions.

    WIRED

All news → · All rights belong to the original creators