What risk is associated with AI systems that function as a black box?

• A. Untrustworthy outputs may lead to poor decisions or exploitation of vulnerabilities.
• B. They always provide transparent reasoning.
• C. They guarantee compliance with all policies.
• D. They reduce security vulnerabilities.

Answer: (A)

Black-box AI systems keep their internal decision paths hidden, so you cannot see how inputs become outputs. This opacity creates a risk that outputs are untrustworthy, biased, or flawed, and you can't validate or audit the reasoning, which can lead to poor decisions or make it easier for adversaries to exploit weaknesses. Because you can't verify why a result occurred, you may miss errors, biases, or vulnerabilities and struggle to enforce safety, fairness, or compliance. The other statements don’t fit: these models do not provide transparent reasoning, do not guarantee policy compliance, and do not inherently reduce security vulnerabilities; in fact, the hidden nature can conceal issues and introduce new risks.