Skip to the content
AI and Cybersecurity Learning and Practice Center Test automation Open the Learning Center

AI risk watchWarnings from authoritative sources

International AI Safety Report 2026: malicious use, malfunctions and systemic risks

International bodyLoss of control

Led by Yoshua Bengio with more than 100 experts and an advisory panel from over 30 countries, the report groups risks into malicious use (deepfakes, cyberattacks, biological and chemical threats), malfunctions (including "scenarios in which AI systems operate outside of anyone's control") and systemic risks (labor market disruption, loss of human skills, psychological effects of AI companions). It calls for defense in depth and notes that once open-weight models are released, their weights cannot be recalled.

Why it matters

This is the closest thing to an international consensus on what could go wrong, and it was written before the summer's agent incidents made the "malfunctions" category concrete.

What you can do

  • Use its three categories to structure your own AI risk register.

Sources

See it with the other risks in the Learning Center