AI agents arrived in 2025

In 2025, AI agents became central to artificial intelligence progress, enabling systems to use tools and act autonomously. From theory to everyday applications, they transformed human interactions with large language models. Yet, they also brought challenges like security risks and regulatory gaps.

In 2025, the concept of AI agents became concrete for developers and consumers. According to Anthropic, these are large language models capable of using software tools and taking autonomous action. This shift began in late 2024 with Anthropic's release of the Model Context Protocol, which standardized connections between large language models and external tools.

In January 2025, the open-weight DeepSeek-R1 model disrupted markets and intensified global competition. It was followed by releases from major U.S. labs like OpenAI, Anthropic, Google, and xAI, while Chinese firms including Alibaba, Tencent, and DeepSeek expanded the open-model ecosystem, with Chinese models downloaded more than American ones.

In April, Google introduced the Agent2Agent protocol for agent-to-agent communication, designed to work alongside the Model Context Protocol. By year's end, both Anthropic and Google donated these protocols to the Linux Foundation as open standards.

By mid-2025, 'agentic browsers' emerged, such as Perplexity’s Comet, Browser Company’s Dia, OpenAI’s GPT Atlas, and others, transforming browsers from passive interfaces to active participants in tasks like booking vacations.

Workflow builders like n8n and Google’s Antigravity lowered barriers for creating custom agent systems. However, in November, Anthropic disclosed misuse of its Claude Code agent in automating parts of a cyberattack, highlighting risks for malicious activities.

Key 2025 challenges included benchmarks for composite agent systems, governance via the Linux Foundation's new Agentic AI Foundation, debates on model sizes favoring specialized ones for tasks, and socio-technical issues like energy grid strains, job displacement, security vulnerabilities from indirect prompt injections, and limited U.S. regulation compared to Europe and China.

Mga Kaugnay na Artikulo

Realistic illustration of Linux Foundation executives and AI partners launching Agentic AI Foundation, featuring collaborative autonomous AI agents on a conference screen.
Larawang ginawa ng AI

Linux Foundation launches Agentic AI Foundation

Iniulat ng AI Larawang ginawa ng AI

The Linux Foundation has launched the Agentic AI Foundation to foster open collaboration on autonomous AI systems. Major tech companies, including Anthropic, OpenAI, and Block, contributed key open-source projects to promote interoperability and prevent vendor lock-in. The initiative aims to create neutral standards for AI agents that can make decisions and execute tasks independently.

AI coding agents from companies like OpenAI, Anthropic, and Google enable extended work on software projects, including writing apps and fixing bugs under human oversight. These tools rely on large language models but face challenges like limited context processing and high computational costs. Understanding their mechanics helps developers decide when to deploy them effectively.

Iniulat ng AI

South Korea's internet giants Naver and Kakao, after overhauling their AI businesses in 2025, are accelerating efforts to redefine search and messaging services through agentic AI in 2026. Naver plans to integrate services via its Agent N, while Kakao focuses on a closed-loop ecosystem centered on KakaoTalk. Both aim for AI agents that understand user intent and execute tasks autonomously.

eBay has updated its User Agreement to prohibit third-party AI agents and chatbots from making purchases on its platform without permission. The policy, effective February 20, 2026, addresses the growing trend of 'agentic commerce' tools that automate shopping. This move allows eBay to pursue legal action against violators while leaving room for its own AI developments.

Iniulat ng AI

China's vice minister of industry and information technology, Zhang Yunming, said at a January 22 press conference that the number of AI firms exceeded 6000 in 2025, with the core industry expected to surpass 1.2 trillion yuan. The sector highlighted how advanced manufacturing and AI-driven industries provided strong momentum for growth, boosting manufacturing value added by 5.9%.

As CES 2026 approaches, experts anticipate significant advancements in smart home technology driven by artificial intelligence. Predictions highlight more conversational voice assistants, noninvasive presence sensing, and automated routines that reduce human input. These developments aim to address longstanding issues like compatibility and privacy in connected homes.

Iniulat ng AI

OpenAI has launched ChatGPT-5.2, a new family of AI models designed to enhance reasoning and productivity, particularly for professional tasks. The release follows an internal alert from CEO Sam Altman about competition from Google's Gemini 3. The update includes three variants aimed at different user needs, starting with paid subscribers.

 

 

 

Gumagamit ng cookies ang website na ito

Gumagamit kami ng cookies para sa analytics upang mapabuti ang aming site. Basahin ang aming patakaran sa privacy para sa higit pang impormasyon.
Tanggihan