Overview and Analysis of Transformative AI Misuse

Executive summary

This report provides an overview and analysis of existential risks posed by the misuse of transformative AI (TAI) technologies. It draws on a 2019 expert survey and presents illustrative disaster scenarios covering a wide range of possible outcomes:

• Decay: Scenarios where AI increases civilizational fragility, potentially leading to extinction or destructive value drift, without direct human extinction or AI takeover occurring.
• Totalitarianism: Scenarios where AI helps a bad value system gain absolute control over most or all of humanity.
• War: Wars caused or worsened by AI technologies lead to global catastrophe or human extinction.
• Non-Anthropocentric Disasters: Scenarios where the future is not dystopian by superficial lights, but there is some great loss of value compared to an ideal (outside the scope of this report).

@online{martin-overview-TAI-misuse-2024,

title = {Overview and Analysis of Transformative AI Misuse},

author = {Samuel Martin},

url = {https://longtermrisk.org/media/Overview-and-Analysis-of-Transformative-AI-Misuse1-10.pdf},

year  = {2024},

date = {2024-10-14},

keywords = {},

pubstate = {published},

tppubtype = {online}

}