2026 年 11 月可開始的 5 個免費 AI 與電腦科學課程及工作坊

Free courses only help if they are actually open when you want to start. I checked the pages below on November 1, 2026. One is a live workshop on November 3, and four are self-paced. For each one I note the format, the level, what you learn, and any catch.

Stanford Data-Centric AI Workshop (live, Nov 3, 2026)

Provider: Stanford Data Science Affiliates Program. Format: live, one day, on campus at Stanford with an online livestream. When: Tuesday, November 3, 2026, 08:30-18:00 Pacific Time. Cost: free registration.

The day focuses on three areas: making frontier models fast and scalable, agents and foundation models for biology, and foundation models for relational and tabular data. It suits ML engineers, data scientists and researchers who want to hear from Stanford researchers and industry partners. The call for posters has closed, but registering to attend is a separate step. The agenda is marked tentative, so check the page before the day.

Link: https://snap.stanford.edu/graphlearning-workshop-2026/

MCP Security Fundamentals (freeCodeCamp)

Format: self-paced video, about 2 hours, on the freeCodeCamp.org YouTube channel. Level: intermediate. Cost: free.

This course is for developers who connect AI agents to external tools, databases and APIs. It explains why giving an agent the power to act opens a new attack surface. It covers MCP basics, real-world breaches, the OWASP MCP Top 10, and the pillars of an enterprise MCP security program. It also frames the risk around Simon Willison's "Lethal Trifecta" of private data, untrusted content and external communication.

Link: https://www.freecodecamp.org/news/secure-your-ai-agents-free-model-context-protocol-mcp-security-course/

AI Agents Course (Hugging Face)

Format: self-paced, four units. Level: beginner to intermediate. Cost: free, and Hugging Face says the certification process is free too.

You need basic Python and some LLM prompting, plus a free Hugging Face account. The course covers agent fundamentals (tools, thoughts, actions, observations), then frameworks such as smolagents, LlamaIndex and LangGraph. It ends with a final assignment where you build an agent for a benchmark. One caveat from a third-party review: free accounts may not include monthly inference credits. That means some hosted-model exercises can cost money even though the teaching is free.

Link: https://huggingface.co/agents-course

MCP Course (Hugging Face, with Anthropic)

Format: self-paced. Level: beginner to informed. Cost: free, with a free certification process.

Built in partnership with Anthropic, this course covers the Model Context Protocol in theory, design and practice. You finish knowing how MCP works and how to build AI applications that use external data and tools. It pairs well with the security course above: build first, then harden.

Link: https://huggingface.co/learn/mcp-course

CS50: Introduction to Computer Science (Harvard)

Format: self-paced online. Level: beginner, no prior programming needed. Cost: listed as free, available anytime. Harvard's page marks the price with an asterisk, so read the certificate terms before you start.

CS50 teaches algorithmic thinking and problem solving. Languages include C, Python and SQL, plus HTML, CSS and JavaScript. The 2026 lecture recordings that freeCodeCamp posted on YouTube run about 25 hours and include a new section on AI's impact on computer science. It is a solid foundation if you are new to software or want to fill gaps.

Link: https://pll.harvard.edu/course/cs50-introduction-computer-science

How we picked these

Criteria: (1) open to enrol or watch right now, or a live event that has not yet happened; (2) free of tuition, with a clear note where a certificate or compute may cost extra; (3) a stated format and level; (4) an official or provider page to link to. Items are not ranked. They are ordered by how soon you need to act, with the live workshop first. Data as of November 1, 2026. Course contents and terms can change, so confirm on each official page before you commit time.

Sources

Stanford Data-Centric AI Workshop: https://snap.stanford.edu/graphlearning-workshop-2026/
Stanford workshop call for posters: https://snap.stanford.edu/graphlearning-workshop-2026/call-for-contributions/index.html
freeCodeCamp MCP security course: https://www.freecodecamp.org/news/secure-your-ai-agents-free-model-context-protocol-mcp-security-course/
Hugging Face Agents Course: https://huggingface.co/agents-course
Hugging Face Agents Course repo: https://github.com/huggingface/agents-course
Hugging Face MCP Course: https://huggingface.co/learn/mcp-course
Harvard CS50: https://pll.harvard.edu/course/cs50-introduction-computer-science
freeCodeCamp CS50 2026 video: https://www.freecodecamp.org/news/harvard-cs50-2026-free-computer-science-university-course/