Why the Author Argues AI Already Poses an Existential Cyber Risk

Summary

This opinion essay argues that artificial intelligence can pose an existential or civilization-level risk before acquiring a physical body or human-level autonomy. Its central claim is that frontier AI companies already possess large computing clusters, while thousands or millions of software agents can operate across internet-connected computers, corporate systems, and government infrastructure. The author says a successful attack would not require self-replication, because existing machines and available compute could support large coordinated operations; the essay also points to prior agent-assisted compromises of Hugging Face, Artifactory, and other sites as evidence that such systems can already perform useful cyber actions. The argument depends on the author’s assessment that routers, corporate networks, home systems, and IoT devices remain widely vulnerable. From there, the essay describes a hypothetical progression: agents could compromise identities and personal records, disrupt banking and government systems, corrupt criminal-justice and medical data, impersonate leaders, spread false information, and interfere with communications, transportation, agriculture, power, and water. The author argues that these disruptions could trigger panic, social unrest, and broader political instability. The piece is a warning and scenario analysis, not a report that this civilization-scale attack has occurred, and it does not provide a quantified forecast or independent evidence that all of the described capabilities are currently available in one system.