人工智能 on IT News

Claude Mythos 5修复

Wed, 01 Jul 2026 14:00:00 +0900

On June 30, 2026, the US Commerce Department lifted export controls on Anthropic’s flagship AI models, Claude Mythos 5 and Claude Fable 5. As of July 1, Mythos 5 has been restored for approved US organizations, while Fable 5 has resumed global availability. This article examines the events leading to this moment and what it means for the AI industry.

The Trigger: Fable 5 Vulnerability Identification

Launched on June 9, Claude Fable 5 and Mythos 5 represented Anthropic’s most advanced AI models to date. Fable 5, in particular, was released to the general public and quickly gained attention for its exceptional coding assistance and creative writing capabilities.

Anthropic Fable 5/Mythos 5恢复

Tue, 30 Jun 2026 14:00:00 +0900

On June 30, 2026, the US Commerce Department lifted the emergency export controls on Anthropic’s Claude Fable 5 and Claude Mythos 5, ending an 18-day suspension that began on June 12. The restoration follows a comprehensive safety agreement between Anthropic and the US government.

The Agreement

The Commerce Department approved Anthropic’s safety package, noting in a statement that “Anthropic’s cooperative approach and swift action set a new benchmark for AI safety. The standards established here may serve as a template for other AI companies.”

Fable 5暂停两周的影响

Thu, 25 Jun 2026 10:00:00 +0900

Nearly two weeks have passed since the US Commerce Department imposed emergency export controls on Anthropic’s Claude Fable 5 and Claude Mythos 5 on June 12. How has the AI industry been affected by this unprecedented suspension? Here is a mid-point assessment.

The Two-Week Timeline

Anthropic immediately announced it was cooperating closely with the US government to resume service as quickly as possible. In its first weekly update, the company reported completing the enhanced prompt detection filter design and inference-time guardrail architecture. By the second week, it had reached a preliminary agreement with the Commerce Department on a third-party security audit framework.

AI智能体的黎明

Thu, 18 Jun 2026 11:00:00 +0900

In recent years, the evolution of artificial intelligence (AI) has been remarkable, moving beyond conversational chatbots that merely answer questions to AI agents that think and act independently. As of 2026, these autonomous AI agents have moved out of the experimental phase and are beginning to integrate deeply into both business operations and daily life. This article analyzes the basic architecture of AI agents, their key components, real-world use cases, and the challenges they face on the path to widespread adoption.

AI搜索与聊天机器人的对比

Sun, 19 Apr 2026 10:00:00 +0900

AI search and chatbots have become two major options for AI-powered information gathering. While both are based on generative AI technology, they excel in different areas and serve different purposes. This article explains their characteristics and how to choose the right tool for your needs.

Characteristics of AI Search

AI search collects the latest information from the internet in real time and generates answers based on the results. Representative services include ChatGPT’s web browsing feature, Perplexity, and Google’s AI Overviews. AI search excels when you need up-to-date news or information from specific web pages.

AI智能体如何改变我们的工作方式

Thu, 26 Mar 2026 10:00:00 +0900

AI agents are evolving rapidly, fundamentally changing how we approach work. Unlike traditional AI that simply provided information or answered basic questions, AI agents can autonomously plan tasks, execute them, and report results in an end-to-end fashion. This article explores how work styles are transforming in the age of AI agents and what we need to do to adapt.

How AI Agents Transform Business Processes

The key feature of AI agents is their ability to create action plans for given goals and execute tasks by selecting the right tools along the way. For example, for meeting minutes, an AI agent can attend the meeting, transcribe the audio, extract key decisions, and automatically email the summary to all participants. This frees humans to focus on review and approval.

生成式AI安全使用规则

Thu, 15 Jan 2026 10:00:00 +0900

As generative AI rapidly spreads across business environments, establishing safe usage rules has become an urgent priority for companies. Allowing unrestricted use without proper safeguards can lead to information leaks and compliance violations. This article covers the fundamental rules for using generative AI safely at work.

Rules for Handling Input Information

The most critical consideration when using generative AI is the information you input. Many generative AI services may use input data for model training. Therefore, directly entering customer information, confidential corporate data, or personal details should be avoided. When using AI for work, consider abstracting or anonymizing information before input.

企业AI采用模式

Sat, 10 Jan 2026 10:00:00 +0900

As more companies pursue AI adoption, the gap between successful and unsuccessful organizations is becoming increasingly clear. The key is not simply deploying the latest AI tools, but selecting an adoption pattern that fits the organization’s characteristics and challenges. This article introduces representative successful AI adoption patterns for enterprises.

Bottom-Up Adoption Pattern

In this pattern, employees in the field voluntarily start using AI tools, and once the benefits are recognized, deployment expands company-wide. The advantages are low initial cost and alignment with actual on-the-ground needs.

在iPhone上使用My Number

Fri, 06 Jun 2025 11:25:47 +0900

Hello! Today, on June 6, 2025, the Digital Agency officially announced that the service "My Number Card on iPhone," which integrates My Number Card functions into iPhones, will start on June 24, 2025. This will allow iPhone users in Japan to use various administrative services through Apple Wallet on their iPhones without carrying a physical My Number Card.

Key Features and Usage

"My Number Card on iPhone" enables users to utilize the following administrative services by adding their My Number Card to Apple Wallet:

DeepSeek开辟的未来——中国诞生的AI革命

Tue, 04 Feb 2025 06:00:00 +0900

Hello! In this post, we will delve into the remarkable China-born artificial intelligence (AI) company “DeepSeek,” examining its background, characteristics, impact on the market, and future outlook. Distinguishing itself from traditional, closed AI development, DeepSeek has drawn significant attention from many developers and researchers by sticking to an open-source stance. Its presence has sent shockwaves through the global AI industry.

DeepSeek’s Background and Founding Journey

DeepSeek was founded in 2023 and has grown rapidly under the leadership of its CEO, Liang Wenfeng, with funding provided by “High-Flyer,” a major Chinese quantitative hedge fund. Mr. Liang, who was previously engaged in AI utilization primarily in the financial sector, launched DeepSeek to take on the challenge of broader AI technology research and development. He has stated that “funding is never the problem, but export restrictions on advanced AI chips are the major challenge.” By spreading open-source models, he aims for the democratization of technology and the improvement of global competitiveness.

DeepSeek开源AI模型趋势

Sat, 25 Jan 2025 09:00:00 +0900

Hello! Today we will analyze the rising open-source AI platform DeepSeek and its rapid adoption across the global AI community. DeepSeek’s emergence has disrupted the conventional wisdom that cutting-edge AI requires massive capital expenditure.

Key Features of DeepSeek

DeepSeek is a high-performance AI model developed by DeepSeek (深度求索), a Chinese AI startup. The model has attracted global attention for several standout characteristics:

Exceptional Cost-Performance: DeepSeek achieved benchmark-comparable or superior results to leading Western models at a fraction of the training cost—reported to be under $6 million compared to the hundreds of millions spent by competitors. This efficiency is achieved through innovative architecture choices including Mixture-of-Experts (MoE) design and advanced optimization techniques.
Strong Coding Capabilities: DeepSeek performs exceptionally well in code logic parsing, syntax generation, and debugging tasks, rivaling specialized coding models. Its performance on competitive programming benchmarks has impressed developers worldwide.
Open-Source Approach: Unlike many commercial AI labs that keep their weights proprietary, DeepSeek has released multiple model variants to the open-source community, enabling researchers and developers to study, fine-tune, and deploy the models freely.

Impact on the AI Landscape

DeepSeek’s emergence has sparked intense debate about AI development strategy. The model demonstrates that strategic innovation in architecture and training methodology can overcome resource disadvantages. This has implications for how companies and countries approach AI investment—potentially democratizing access to state-of-the-art AI capabilities.

ChatGPT商务使用效率技巧

Sun, 19 Jan 2025 09:00:00 +0900

Hello! Today we will cover some practical tips to use ChatGPT at work to boost your daily efficiency. Whether you are in marketing, engineering, or operations, AI tools can dramatically reduce time spent on routine tasks.

Key Use Cases

Summarization and Proofreading: Upload lengthy documents, reports, or meeting transcripts and ask ChatGPT to extract key points, identify action items, or review grammar and clarity. Save hours of manual reading and editing time.
Email Generation: Draft professional business emails from brief bullet points or voice memos. Specify the tone (formal, friendly, urgent) and ChatGPT can produce polished correspondence in seconds, adapting to the recipient and context.
Brainstorming Partner: Generate creative copywriting ideas, marketing campaign concepts, product names, or project approaches. By prompting ChatGPT with relevant constraints and target audience details, you can rapidly explore dozens of possibilities before refining the best ones.
Code Assistance: Developers can use ChatGPT to debug code snippets, generate boilerplate functions, explain unfamiliar syntax, or translate code between programming languages. This accelerates development and reduces context-switching.
Data Analysis: Upload CSV files or paste structured data and ask ChatGPT to identify trends, generate summary statistics, or create visualizations through Python code that can be run locally.

Tips for Better Prompts

Simple rules to get high-quality outputs from AI:

AI与Apple的合作

Sat, 04 Jan 2025 07:00:00 +0900

Hello! In this post, we will look into the latest AI assistant technology being developed by Apple, examining its impact on daily life and future possibilities. Apple products are widely recognized for their ease of use and innovative technology, but recently, the evolution of artificial intelligence (AI) has drawn even more attention. Based on the latest trends, let’s explore in detail how next-generation AI assistants will transform our lives.

Evolution of Apple’s AI Assistants

When talking about Apple’s AI, many people probably think of the voice assistant Siri. Since its debut in 2011, Siri has continued to expand its capabilities. Recently, there have been rumors that Apple is working to integrate new conversational AI technologies into Siri. For example, the following enhancements are expected: