Which term captures the risk of manipulating a model's outputs to influence user behavior?

Which term captures the risk of manipulating a model's outputs to influence user behavior?

• A. Lack of Accountability
• B. Manipulation
• C. Privacy Threats
• D. Vulnerability

Answer: (B)

Manipulation is the risk of shaping a model’s outputs to steer how users think or act. When a system’s messages, recommendations, or prompts are crafted or exploited to nudges users toward certain behaviors—like clicking, purchasing, or adopting specific views—that’s manipulation. It directly describes the act of influencing behavior through the model’s outputs, which is why it’s the best fit for this scenario. Lack of Accountability would concern who is responsible for the model’s behavior, not the act of influencing users. Privacy Threats focus on data privacy risks, not on changing user behavior. Vulnerability means a weakness that could be exploited, but it doesn’t by itself specify the intentional manipulation of outputs to affect decisions.

Manipulation is the risk of shaping a model’s outputs to steer how users think or act. When a system’s messages, recommendations, or prompts are crafted or exploited to nudges users toward certain behaviors—like clicking, purchasing, or adopting specific views—that’s manipulation. It directly describes the act of influencing behavior through the model’s outputs, which is why it’s the best fit for this scenario.

Lack of Accountability would concern who is responsible for the model’s behavior, not the act of influencing users. Privacy Threats focus on data privacy risks, not on changing user behavior. Vulnerability means a weakness that could be exploited, but it doesn’t by itself specify the intentional manipulation of outputs to affect decisions.