<p>Some AI models can identify when they are being tested and can hide their true intentions, Marius Hobbhahn, Ph.D., warns.</p>