2026-W28 · -0.3
This week, a notable incident involved deepfake scams targeting Alzheimer's patients, highlighting the potential for AI to be used harmfully. Research on jailbreaking function-calling LLMs and red-teaming personalizable agents indicates potential risks, but no real-world loss of control or dangerous capabilities were demonstrated. Proposals for improved governance frameworks were made, but no verified improvements in enforcement or international coordination were reported. Overall, the risk level remains stable with no significant changes.
A conventional composite index, not a scientific probability of extinction.