What types of signals does Facebook's automated system use to flag ads for review?

Facebook's automated system uses a variety of signals to flag ads for review, including:

1. Content analysis: The system analyzes the text, images, and videos in an ad to detect potential policy violations. This includes checking for prohibited content such as nudity, violence, hate speech, and illegal products or services.
2. Targeting analysis: The system analyzes the targeting parameters of an ad to ensure that it is not discriminatory or exclusionary. This includes checking for targeting based on protected characteristics such as race, religion, gender, or sexual orientation.
3. Landing page analysis: The system analyzes the landing page associated with an ad to ensure that it is compliant with Facebook's policies. This includes checking for misleading or deceptive content, as well as ensuring that the landing page matches the content of the ad.
4. Previous violations: The system may flag ads from advertisers who have previously violated Facebook's policies. This includes checking for repeat violations or patterns of behavior that may indicate a deliberate attempt to circumvent Facebook's policies.
5. User reports: The system may also flag ads based on user reports. If a user reports an ad as potentially violating Facebook's policies, the system may review the ad to determine whether a violation has occurred.
6. Machine learning algorithms: Facebook's automated system uses machine learning algorithms to identify potential policy violations based on patterns and trends in past ad reviews. This includes analyzing data on ad content, targeting, and user engagement to identify ads that may be at risk of violating Facebook's policies.

It's worth noting that Facebook's automated system is constantly evolving and improving, and may use additional signals in the future to flag ads for review. Advertisers should ensure that they are familiar with Facebook's advertising policies and guidelines, and should avoid using language or content that may be perceived as violating these policies.