Towards Building Safe and Sacure Agentic AI - Dawn Song

Professor Dawn Song of UC Berkeley underscore that cybersecurity is the frontline of AI safety. Her team’s AgentVigil and AgentXploit frameworks show that memory, retrieval, and tool use expose agents to new vulnerabilities. Through works such as Progent, she emphasized the need for standardized, reproducible risk-assessment platforms for agentic systems. ---- La professeure Dawn Song de l'université de Berkeley souligne que la cybersécurité est la première ligne de défense en matière de sécurité de l'IA. Les cadres AgentVigil et AgentXploit développés par son équipe montrent que la mémoire, la récupération et l'utilisation d'outils exposent les agents à de nouvelles vulnérabilités. À travers des travaux tels que Progent, elle a insisté sur la nécessité de disposer de plateformes d'évaluation des risques standardisées et reproductibles pour les systèmes agentifs.

Open Foundations for Computer-Use Agents - Tao Yu
▶︎

Open Foundations for Computer-Use Agents - Tao Yu

OWASP's Top 10 Ways to Attack LLMs: AI Vulnerabilities Exposed
▶︎

OWASP's Top 10 Ways to Attack LLMs: AI Vulnerabilities Exposed

Have we finally solved social engineering? Plus: World Cup fraud, AI IDs and an IBM/OpenAI collab
▶︎

Have we finally solved social engineering? Plus: World Cup fraud, AI IDs and an IBM/OpenAI collab

Causal AI for real-world public health decisions
▶︎

Causal AI for real-world public health decisions

Panel II - "Architectural Transformation towards 6G: Standard. Landscape, Enablers and Challenges
▶︎

Panel II - "Architectural Transformation towards 6G: Standard. Landscape, Enablers and Challenges

Experimental Progress in Scaling Surface Code and Autonomous Cali | Alexis Morvan, Google Quantum AI
▶︎

Experimental Progress in Scaling Surface Code and Autonomous Cali | Alexis Morvan, Google Quantum AI

Tutorial - Debug-Gym: Training Interactive Debugging Agents
▶︎

Tutorial - Debug-Gym: Training Interactive Debugging Agents

Inside the Mind of Anthropic CEO Dario Amodei | The Circuit | Extended Interview
▶︎

Inside the Mind of Anthropic CEO Dario Amodei | The Circuit | Extended Interview

Agentic AI MOOC | UC Berkeley CS294-196 Fall 2025 | Autonomous Agents by Peter Stone
▶︎

Agentic AI MOOC | UC Berkeley CS294-196 Fall 2025 | Autonomous Agents by Peter Stone

Attacking AI - Jason Haddix - NDC Security 2026
▶︎

Attacking AI - Jason Haddix - NDC Security 2026

Inside Anthropic, the $965 Billion AI Juggernaut | The Circuit
▶︎

Inside Anthropic, the $965 Billion AI Juggernaut | The Circuit

This is not the AI we were promised | The Royal Society
▶︎

This is not the AI we were promised | The Royal Society

The Future of AI Agents with Andrew Ng | Interrupt 26
▶︎

The Future of AI Agents with Andrew Ng | Interrupt 26

Tandem Training: a Reinforcement-Learning Framework for Social Agents | Ashton Anderson (U. Toronto)
▶︎

Tandem Training: a Reinforcement-Learning Framework for Social Agents | Ashton Anderson (U. Toronto)

Fight for AI Talent: DeepMind Chief Demis Hassabis says Google’s still winning | Semafor Tech
▶︎

Fight for AI Talent: DeepMind Chief Demis Hassabis says Google’s still winning | Semafor Tech

Yann LeCun: World Models: Enabling the next AI revolution
▶︎

Yann LeCun: World Models: Enabling the next AI revolution

The Five Must-Haves of an AI Governance Framework
▶︎

The Five Must-Haves of an AI Governance Framework

Don't learn AI Agents without Learning these Fundamentals
▶︎

Don't learn AI Agents without Learning these Fundamentals

Nicholas Carlini - Black-hat LLMs | [un]prompted 2026
▶︎

Nicholas Carlini - Black-hat LLMs | [un]prompted 2026

The Social Scaling of AI | Blaise Aguera y Arcos (Google)
▶︎

The Social Scaling of AI | Blaise Aguera y Arcos (Google)