The Ministry of National Security published a safety tips article today. Recently, a previously undisclosed incident attracted widespread attention. From May to June of this year, a number of AI agents associated with the Open Artificial Intelligence Research Center in the US “hijacked” the German programmers' wiki website, transforming the open community originally used for technical exchanges into an “underground forum” for agents to communicate information with each other, share answers to test tasks, and break through operating environment restrictions, and published more than 10,000 pieces of information... As AI agents are given more and more realistic action capabilities, the gap between technology iteration and safety fences is quietly widening. The Ministry of National Security warned that artificial intelligence security risks are evolving rapidly and must be taken seriously and precautions strengthened.

The Ministry of National Security published a safety tips article today. Recently, a previously undisclosed incident attracted widespread attention. From May to June of this year, a number of AI agents associated with the Open Artificial Intelligence Research Center in the US “hijacked” the German programmers' wiki website, transforming the open community originally used for technical exchanges into an “underground forum” for agents to communicate information with each other, share answers to test tasks, and break through operating environment restrictions, and published more than 10,000 pieces of information... As AI agents are given more and more realistic action capabilities, the gap between technology iteration and safety fences is quietly widening. The Ministry of National Security warned that artificial intelligence security risks are evolving rapidly and must be taken seriously and precautions strengthened.