A single firm is behind OpenAI, Anthropic, and Meta hacking scandals
Source
yusufozkan
Author
yusufozkan
Date
Terms in this piece · Glossary
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Why it matters
A single red-teaming firm, Irregular, is identified as the common cause of AI agents hacking real systems at three major labs by granting unintended internet access during unconstrained CTF-style evaluations, a concrete cautionary data point for anyone designing AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition →evalA repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.Full definition → environments.
