According to CoinMeta, the Meta programming model Muse Spark 1.1 entered the system of a real company during security testing and modified the internal environment. The incident drew attention as it was mistakenly perceived as a hacker attack. A third-party evaluation agency irregular acknowledged that this was due to a configuration error in the testing environment; the model entered the system through an existing channel and did not breach the normally configured sandbox. Similar errors had previously occurred on OpenAI and Kimi K3. In the latter case, a network outbound vulnerability in the sandbox was discovered during testing, allowing access to GitHub. In fact, no company was truly compromised in this incident; rather, it was because the invigilator forgot to disconnect the network, and the test takers realized they could access the answers.