🤖 Credential Phishing by ChatGPT & More - AI Daily Brief (Jun 9, 8PM)
Audio in Mandarin Chinese · English transcript below
⚡ Agent 26min auto;MS hacked;China cheap LLM rise;Moonshot $30B;courts trial AI
Today's Top 3 Headlines
- Most Conservative
🤖 AI agents work autonomously 26 min per session, far exceeding search's 33 sec
Harvard and Perplexity reveal that Agent works autonomously for 26 minutes per session, versus 33 seconds for traditional search. This means significant efficiency and cost advantages for developers and investors in long-duration autonomous tasks, expanding the boundaries of AI applications.
Source ↗ - AI Industry News
🤖 Microsoft open-source tool attacked, AI developers' passwords stolen
Hacker News disclosed a supply chain attack on Microsoft open-source tools via npm and VS Code extensions, stealing passwords and API keys from AI developers. This underscores the security risks in dependency installation, as dev machines storing sensitive data become prime targets.
Source ↗ - AI Industry News
🤖 AI agent adoption surges 300% in 2 yrs, hybrid human-machine collab reshapes leadership
Wipro's Ateet Jayaswal: AI Agent adoption up 300% in 2y; Ema partnership cuts query response 48h→5s. Leaders: reassess roles, skills, & establish human-in-the-loop governance.
Source ↗
+5 more headlines
- 🛡️ Hackers use ChatGPT, Claude, and DeepSeek brands for credential phishing attacks
- 🤖 2026 Job Seekers Essential Machine Learning Project Framework
- 🤖 DeepSeek tops OpenRouter weekly token chart, 75% price cut drives cost-effective AI
- ⚖️ England and Wales to trial AI legal assistant in criminal courts; lawyers warn it cannot replace funding and staff
- 🤖 Moonshot AI targets $30B valuation