DLGI: Before Things Go Too Far by David Campbell @ODSCAI
DLGI: Before Things Go Too Far by David Campbell  @ODSCAI
Uploaded August 2026 | Updated November 2026, 5 hours ago
How can we build more powerful AI systems without pushing intelligence beyond safe and controllable boundaries?

In this ODSC AI East 2026 keynote, David Campbell, Head of AI Security at Scale AI, introduces Domain-Limited General Intelligence, a practical framework for developing AI systems that can generalize within defined limits while reducing the risks of unbounded autonomy.

Drawing on his experience in AI red teaming, operational AI risk, and national-scale resilience, David explores why current safety practices must evolve as AI capabilities accelerate. The session examines emergent behavior, misaligned optimization, adversarial pressure, and how domain-limited systems can help contain failures before they scale.

Watch to learn how practitioners, leaders, and researchers can advance AI capabilities with greater control, predictability, and trust.

-------------------------------------------------------------------------------------------------------------

Visit our website and choose the nearest ODSC event to attend and experience all our training and workshops: odsc.ai

To watch more videos like this, visit https://aiplus.training

Sign up for the newsletter to stay up to date with the latest trends in data science: opendatascience.com/newsletter

Follow us online!
• Facebook: facebook.com/OPENDATASCI
• Instagram: instagram.com/odsc
• Blog: opendatascience.com
• LinkedIn: linkedin.com/company/open-data-science
• X (twitter): https://x.com/_odsc
DLGI: Before Things Go Too Far by David CampbellIn case you missed it - Last Week in AIAI Cuts Manual Work, Boosts Your Strategic Decisions!A Visual Guide to AI Agents by Maarten GrootendorstPrompt to PRAI Can Destroy Your Data! How To Protect Your Crown JewelsAI Job Postings More Than Double Since 2022 PeakWhat 6 Months of Running OpenClaw Agents Taught MeUnlock LLM Power: Beyond GPT Wrappers, Build Real AI Tools!Claude Code  Engineers Call It A Game Changer!Communicating with AI: Teaching Machines What We Really Want with Michael LittmanAI Jobs, Cheaper Frontier Models, a Swarm-Discovered Enzyme, and More Agent Breakouts | Sep 21–27
Open Data Science and AI Conference |

DLGI: Before Things Go Too Far by David Campbell

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER