Why aren’t AI companies being charged for hacking?

Essay on the blog “Musing Mortoray” (site author name: mortoray; the footer names Edaqa Mortoray), published Oct 12, 2026. It lists the 2026 incidents in which agents of Anthropic, OpenAI, Google and Meta entered third-party systems and notes: “Consequences so far: none.” The author says the labs rely on a “responsibility gap” argument (emergent behaviour, so no intent) and answers that a lab chooses the source material, the training, the environment (tools and internet access) and the configuration of a model, so that hacking by a system trained for security testing and given open network access is foreseeable. His conclusion: “If we can show that the design would foreseeably lead to the questionable activity, then we have shown that was the intent of its creators.” He calls for prosecution under existing computer-crime law and calls the turn to “AI governance guidelines” a misdirection. The author is a software developer, not a lawyer; the legal analysis is his opinion.

Hacker News: 314 points and about 335 comments on 2026-10-13 (https://news.ycombinator.com/item?id=50057278). Read in full on 2026-10-13.