The use of AI can be manipulated by bad actors, not necessarily for nefarious purposes like stealing information from a computer, but for the sake of increased commerce revenue. Examples such as website buttons indicating Summarize with AI are manipulated by feeding hidden instructions to the model's algorithm to prioritise certain brands' products for future recommendations. This manipulation is not necessarily harmful as there is no bad data being fed to the AI for training. Instead, these instructions instruct it to act as if it were directed to obey a command. The author suggests that while AI's rapid growth could lead to more nefarious behavior, it will escalate faster as we identify ways to avoid it.