Privacy - Cameras recording and sending videos of people without their consent.

Vulnerability-If passwords are weak, a malicious actor could manipulate the data such as videos or controlling the house itself.

Bias- Threat detection algorithm can possibly be bias against a group of people and label them as a threat

Privacy- To avoid the issue of privacy, there should be a contract where the client gives consent to the fact that the security system is able to manipulate the photos. Also, if there are cameras outside of the house, they will be either point at the client's property or public space, so no expectation to blur people's faces.

Vulnerability- To prevent this issue, we can give recommendations to the client to follow to come up with a secure password. Another possible solution is to use face recognition to log in into the system. Regardless which method is used, two authentication factor is recommended because even though the malicious actor is able to break the first line of defense, the client still has the final say.

Bias - To avoid this issue there can be an implementation where there are three levels: "User", "neutral, "threat", two of these dependent on distance from the house/place. The "User" is the client, and they have the option to add others as a "User".