AI Agent News
Track major events, funding, model releases and breakthroughs across the AI Agent landscape
Latest industry news
Track major events, funding, model releases and breakthroughs across the AI Agent landscape
Key events timeline
OpenClaw erupts on GitHub
OpenClaw hits global GitHub Top 10 in 10 days, outpacing the Linux kernel star growth
Meta acquires Manus for $2B
Meta acquires Manus AI for $2B, locking in the general-purpose Agent race
DeepSeek-V3 open-sourced
The value king, at just 5% of GPT-4 cost
Manus goes viral overnight
The world's first general-purpose AI Agent draws unprecedented attention
OpenAI Deep Research
OpenAI ships a deep-research Agent that generates professional reports in one click
MCP Servers pass 500
The MCP ecosystem erupts — 500+ servers built in 3 months
DeepSeek-R1 stuns the world
Open-source reasoning model at just 3% of OpenAI cost, reshaping the global AI landscape
MCP protocol born
Anthropic releases the Model Context Protocol, the de facto standard for Agent interfaces
Claude Computer Use
Anthropic lets AI directly control the computer screen for the first time, opening a new paradigm
Replit Agent full-stack automation
Natural language to a shipped product, aimed at non-engineers
Cursor ARR passes $100M
The fastest-growing SaaS ever, the new king of AI coding tools
Claude 3.5 tops SWE-bench
The strongest coding AI, bug-fixing at a junior engineer level
Devin launches
The world's first autonomous AI software engineer, able to complete full coding tasks on its own
AI Drug Discovery Milestone: First Fully AI-Designed Molecule Passes FDA Phase III Trial
Insilico Medicine's INS018_055 becomes the first candidate drug fully designed by AI throughout the entire drug discovery process, successfully passing FDA Phase III clinical trials, with potential approval by the end of 2026. Meanwhile, AI-discovered drugs from Recursion Pharmaceuticals and Exscientia have also entered late-stage clinical trials. This article analyzes the technological breakthroughs in AI-driven drug discovery, estimates how AI will reshape the trillion-dollar pharmaceutical market, and highlights AI drug discovery targets that investors are watching.
AI Ethics Dilemma 2026: How Bias, Discrimination, and Hallucinations Affect Real Users
A joint 2026 AI ethics report from MIT and Stanford reveals that mainstream large language models still exhibit significant bias in race- and gender-related tasks, with medical diagnostic AI showing 23% lower accuracy for people of color and hiring AI systematically scoring female resumes 15% lower. This article examines the sources of AI bias, current mitigation measures, and how users can identify and avoid bias risks when using AI tools in daily life.
Advances in Large Model Hallucination Research: New Techniques Reduce Hallucination Rate by 60%
Stanford University and Anthropic jointly released research on hallucination mitigation, combining retrieval augmentation, self-consistency checks, and uncertainty quantification to reduce the hallucination rate on the TruthfulQA benchmark from 35% to 14%. The study found that hallucinations mainly occur near the model's "knowledge boundary"; actively expressing uncertainty is a trainable capability; and RAG combined with citation tracking yields the best results.
China AI Talent Report: Algorithm Engineer Shortage Exceeds 500,000, Salaries Up 35% Year-on-Year
Liepin and Tsinghua University released the 2025 China AI Talent Report. It shows that China's AI professional talent gap has exceeded 500,000, with the average salary for AI algorithm engineers reaching 350,000-600,000 RMB per year (3-5 years of experience), up 35% year-on-year. Talent in large models is particularly scarce, with top large model engineers earning annual salaries of 2-4 million RMB. AI enrollment in universities has tripled, but still falls far short of demand.
AI Energy Crisis: Data Centers to Consume More Electricity Than France in 2025
The IEA predicts that global AI data centers will consume 600 TWh of electricity in 2025, about 1.2 times France's total electricity usage. A single large model training run consumes enough electricity to power 1,000 homes for a year. Major AI companies have pledged to achieve 100% renewable energy by 2030, but coal power demand will still increase in the short term. Improving AI chip efficiency is a key solution.
AI Coding Agent Breaks 70% on SWE-bench: Software Engineering Enters Semi-Automation Era
Multiple companies have achieved major breakthroughs on the SWE-bench Verified benchmark (real GitHub issue fixes): Claude 3.7 Sonnet reached 62.3%, Devin 2.0 hit 67.5%, and an unnamed startup's agent reached 71.8%. This means AI can now reliably complete over 60% of real-world software engineering tasks, marking the transition from 'AI-assisted' to 'AI-led specific tasks' in software engineering.
2030 AI Forecast Report: Which Jobs Will Disappear, Which Will Emerge, and Which Cannot Be Replaced
The McKinsey Global Institute and AI Now Institute jointly released a report on AI's impact on employment by 2030. It predicts that AI will significantly alter 12% of global jobs by 2030, while also creating new roles such as AI trainers, human-machine collaboration specialists, and AI auditors. Data processing and basic cognitive tasks face the highest risk; creativity, emotional intelligence, and complex physical manipulation are the hardest to replace.
Google DeepMind Discovers 3 New Theorems in Mathematics with Gemini 2.0
Google DeepMind announced that its Gemini 2.0 model, with assistance from mathematicians, discovered three previously unknown theorems in combinatorics and topology, providing complete proofs. This marks the first time AI has genuinely advanced the frontier of pure mathematics (not just verifying known theorems), hailed by Nature as "a new milestone in AI-driven scientific discovery."
Anthropic Interpretability Breakthrough: First Direct Reading of Claude's 'Thoughts'
Anthropic research team published a paper claiming they can partially directly read concept representations in Claude's 'brain'. By decomposing intermediate activations using sparse autoencoders, they identified features related to emotions like 'fear' and 'gratitude'. This breakthrough has significant implications for AI safety (verifying model's true objectives) and AI welfare (understanding whether AI has internal states).