Debate: Are LLMs Hacking Tools or Superhuman Attack Swarms?

joshua_saxe · x · 2026-09-01

There is a divide in the security community regarding the capabilities of LLMs and Agents. One view argues they are specific implementations of known concepts, not sci-fi terminators. The other counters that current models far surpass past capabilities like cat/dog classification, enabling "superhuman hacking swarms" that perform real-world pwning of companies.

Original post →

More from Safety

Safety channel →