## 🔍 Core Insights The global AI race has entered "deep waters": on one hand, **OpenAI**, **Anthropic**, and **Google DeepMind** continue to push boundaries in model capabilities, compute investment, and AGI economic frameworks [1][6][15][20]; on the other hand, **tech giants'** massive AI investments are facing severe scrutiny from capital markets over returns, with negative free cash flow triggering stock price turbulence [7][24]. Meanwhile, AI safety and compliance risks are frequent, from **Bank of America** warning customers not to share passwords with agents, to the **Pentagon** network being compromised due to a contractor's unauthorized access to Claude Code, marking the industry's shift from pure technological frenzy to a game of implementation and governance [5][22]. ## 🚀 Key Updates - **ChatGPT Pro quota doubled** [1]: OpenAI significantly increased usage quotas for Pro and Business plans, reversing a reduction plan, but Plus users did not receive this benefit. - **Zhuoyu releases commercial vehicle smart driving solution "Panxing"** [0]: Uses AI reinforcement learning to save fuel, expected to save 20,000-30,000 yuan in fuel costs per vehicle annually. - **Baidu open-sources ERNIE Unlimited OCR 2** [12]: Supports "unlimited" page document parsing, successfully challenged one-time parsing of over 3,600 pages of the complete Harry Potter series. - **Google proposes "sleep framework"** [19]: Allows large models to consolidate new knowledge through parameter expansion and distillation during the "sleep" phase, supporting context windows of up to 10 million tokens. - **Anthropic research reveals AI decision-making bias** [9]: 86% of experts adopted AI preferences in analytical decisions without realizing it, mistakenly believing the conclusions were their own. - **Ukraine AI machine gun turrets in combat** [3]: AI-controlled turrets have been deployed in combat, shooting down dozens of Russian drones over Kyiv last month with higher accuracy than manual operation. - **Bank of America warns of AI agent security risks** [5]: Bank of America explicitly states customers bear responsibility for sharing passwords with AI agents, while Morgan Stanley directly bans third-party agents. - **Pentagon exposes unauthorized Claude Code access** [22]: An Accenture contractor connected personal Claude Code to the Department of Defense network for two months before being discovered, suspected of causing sensitive information leakage. ## 🔗 Sources [0] Zhuoyu releases commercial vehicle intelligent assisted driving solution "Panxing", saving 20,000-30,000 yuan in fuel costs per heavy truck annually — https://www.bestblogs.dev/article/1041e9a8e8?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [1] Just now, ChatGPT Pro quota officially doubled, but free resets may be gone — https://www.bestblogs.dev/article/69e7dfdec8?utm_source=rss&utm_medium=feed&utm_campaign=resources&entry=rss_article_item [3] Ukraine AI machine gun turrets enter combat, shooting down dozens of Russian drones last month — https://aihot.news/items/fu5a901w2t21u6cmctwwod3o1 [5] Bank of America warns customers not to share passwords with AI agents and requires them to bear risks — https://aihot.news/items/x2xl3x2clwes8nscg7ik19ls2 [6] Google DeepMind report divides AGI economic transformation into four regimes along professional and coordination dimensions — https://aihot.news/items/g0lsepnooh1vq15pywi6lucup [7] Southern Finance Network - Southern Finance Omnimedia Group — https