What academic research adds to dating-safety advice.

What it adds: It helps explain why a private-image request is a safety decision involving pressure, consent, privacy, and possible later harm, not merely a question of whether a profile looks real.

Used here for: Private-photo, sexting, image-sharing, and “prove you trust me” guidance.

Limit: These studies do not provide an all-age prevalence estimate or a person-level detector. A pressured request is a reason to pause, not proof of criminal intent.

What it adds: It puts disclosure and informed consent on the safety map: polished language or a fast emotional connection may be shaped by a tool, and that possibility cannot be settled by a verification badge.

Used here for: Verification limits, AI-assisted profiles and messages, and the difference between a presentation signal and a person’s underlying story.

Limit: The study does not show that AI use causes harm, that undisclosed AI use is always deceptive, or that a particular person used AI.

What it adds: It prevents a common mistake: treating safety as only the user’s responsibility. A feature can help in one moment while also creating privacy, policing, or exclusion concerns for another user.

Used here for: Report and support guidance, privacy boundaries, and the section explaining why we do not rank platforms or promise that a feature prevents harm.

Limit: This is not an app ranking, an incident-rate comparison, or proof that every platform feature has the same effect in every community.

What it adds: It offers a useful information architecture: safety is a route through several moments and support options, not a single badge or a final verdict.

Used here for: The report’s stage-based risk path, community context, and links to support or reporting routes.

Limit: The existence of a map or guide is not an outcome evaluation. It does not prove that a user will be safe or that one route fits every location.

What it adds: It helps describe the transition from app chat to another channel or an in-person meeting as a change in context, not as an automatic sign of danger.

Used here for: The “common risk path” sequence and advice to reassess privacy, pressure, and reversibility at each stage.

Limit: This is not a romance-scam prevalence study and does not establish a universal safe timeline for moving off-app or meeting.

What it adds: It keeps emerging AI-enabled manipulation and family or community support in view without treating an older person’s age as a diagnosis of vulnerability.

Used here for: Emerging-threat context and advice about involving a trusted person when pressure, isolation, or money requests make decisions harder.

Limit: This announcement is not a completed peer-reviewed finding. It must not be used to claim that older adults are uniformly more likely to be deceived.