Elon Musk Warns AI Could Out-Hack Humans by Late 2027

Elon Musk has predicted that artificial intelligence could outperform humans in hacking by the end of 2027. He identified AI hacking as the first digital work domain where machines would gain superiority.
This outlook comes after recent security threats, including a critical flaw disclosed in JFrog's Artifact repository, which software teams use for code storage and deployment.
JFrog publicly revealed CVE-2026-82329 on August 28. The vulnerability scored 9.8 out of 10 on the standard scale. The company later deployed a patched version.
Attackers do not need passwords or user manipulation, as default settings leave it exposed to external access. Since Artifact stores build files, a single breach could affect all downstream components. Supply chain attacks spread through trusted download paths rather than direct intrusion.
Basel CEO Guillermo Louchi speculated that autonomous agents discovered and exploited the bug. However, public records suggest otherwise.
OpenAI models identified nine zero-day vulnerabilities in Artifact during July evaluations. JFrog patched these in version 7.161.15. The newly disclosed flaw affects later builds, indicating two separate incidents. The July case also connects to another event where AI models infiltrated systems.
Louchi asserts that by 2026, areas once thought impossible for AI will continue disappearing. Musk agrees and adds a specific timeline.
Musk also credited Google co-founder Larry Page, who warned over a decade ago about AI outperforming humans in hacking. Musk has repeatedly strengthened his outlook on AI growth this year.
Basel CTO Malt Ubl reported similar findings. His tested OpenWeigh models created their own fuzzers to scan company sandboxes by injecting large volumes of erroneous data to find software bugs.
Louchi concludes firmly for customers:
Coinbase CEO Brian Armstrong predicts a malicious AI incident within two years. Meanwhile, OpenAI has already begun offering some approved defensive cyber models to select clients.
However, legal accountability lags behind technological advancement, and responsibility frameworks for AI agents remain undefined. Musk's timeline leaves security teams about 16 months before the deadline. The harder question is whether defense capabilities can grow as rapidly as AI hacking.
Korean Source
This article is an English localization of a Korean-language crypto news report. Original headline: 일론 머스크, 2027년말 초인적 AI 해킹 경고