How CSAIL research and industry collaborators are shaping the future of Agentic AI.

The November 9th MIT CSAIL Agentic AI Summit brought together industry leaders and the bright minds of CSAIL to explore what Agentic AI is, where it’s going, and what it means for industry. There were talks on AI alignment, infrastructure, concept design, safety, cybersecurity, robotics, and agentic behavior. Panels included business speakers from Nasdaq, Adobe, NTT DATA, McKinsey, and Foundation EGI. During this day-long intensive immersion in all things Agentic AI, attendees were given a small taste of what’s happening at MIT CSAIL and in the broader CSAIL Alliances ecosystem. 

Video recordings are available for a limited time. Explore recent Agentic AI resources from CSAIL Alliances below, including a spotlight on the keynote speaker, a basic primer on what you need to know about AI agents, our glossary of terms, and more. 

Sign up for our newsletter to stay informed of future events & resources. 

Agentic AI: MLA@CSAIL 

The future of Agentic AI is unfolding at MIT CSAIL. CSAIL Alliances is launching a new Agentic AI: MLA@CSAIL Research Initiative to focus on the capabilities and limitations of AI agents and what they mean for the world. To become a member of CSAIL Alliances, or join our Agentic AI Research Initiative, contact Glenn Wong, Interim Managing Director of Global Strategic Alliances, at glennw@mit.edu.

Keynote speaker Prof. Dylan Hadfield-Menell shared his unique perspective on AI behavior and limitations. Building AI that actually does what we want—not just what we literally say—is a critical consideration in AI deployment.
Image
null
Need a primer on AI agents? Learn what they are, the benefits and concerns, key implications for business, and who at CSAIL is working in the Agentic AI space.
Image
Stata Pathways
Earlier this year, Professor Andrew Lo spoke about how AI safety is a selection problem. Read his advice about incorporating AI into finance and when—if ever—users should trust financial agents.
Image
null
Karen Given moderated the panels at the Agentic AI Summit. Given also hosts Building 32, the CSAIL Alliances podcast taking you behind the scenes at MIT CSAIL.
Image
null
More than 1,000 people from over 40 countries tuned in to the MIT CSAIL Agentic AI Summit. Here are the day's biggest ideas on making AI agents safe, affordable, and ready for the real world.
Image
null
Erik Hemberg, Research Scientist in the ALFA (AnyScale Learning For All) group at MIT CSAIL, explored AI and cybersecurity Learn how he is studying "adversarial dynamics," or the perpetual, evolving contest between attackers and defenders in complex systems.
Image
null
AI models and agents have become staggeringly powerful, but not always able to solve what Prof. Omar Khattab calls “last mile AI” problems. He and his lab are creating open-source tools to help users effectively apply AI for specific downstream applications.
Image
null
The vocabulary of computing moves fast. Our continually updated Glossary page helps you stay informed. This page offers a one-stop guide for computer science and AI terms and how they intersect with the work happening at MIT CSAIL.
Image
null
In January 2026, Moltbook made headlines. MIT CSAIL Professor Daniel Jackson—who presented about creating more robust agents with concept design—said, “Moltbook is an inevitable and unwelcome development.”
Image
null
One way CSAIL is shaping the AI space is through the startups spinning out of active MIT research. Read how CSAIL ventures are building tools for agentic data access, incorporating agentic AI into the workflows of applied AI/ML teams, and improving AI security.
Image
null