Notifications
Clear all
Topic starter
14/09/2026 10:09 am
TL;DR: Fraudsters can cheaply fabricate social profiles, and existence-based social matches often confirm only that an identity appears online, not that it is trustworthy, according to Sift. With fake accounts, synthetic backstories, and AI-generated personas scaling quickly, behaviour-based and cross-network signals are proving more resilient than surface-level social proof.
NHIMG editorial — based on content published by Sift: Why Social Media Signals Don’t Equal Consumer Trust
By the numbers:
- Facebook removed 1.1 billion fake accounts in the fourth quarter of 2025, up from 698 million in the previous quarter.
- 37% of influencer followers in an independent analysis, t analysis of 100,000 Instagram and TikTok accounts showed signs of being fake, purchased, or otherwise inauthentic.
- Fraud losses reached nearly $1 billion to business impersonators in 2025 alone, according to FTC reporting cited by Sift.
Questions worth separating out
Q: How should fraud teams evaluate social media signals before using them in identity decisions?
A: Fraud teams should treat social media data as one input, not proof of legitimacy.
Q: Why do fake profiles keep passing basic identity checks?
A: Because many checks measure existence rather than consistency.
Q: What signals are better than social proof for detecting synthetic identity fraud?
A: First-party behavioural signals are usually stronger.
Practitioner guidance
- Demote social profile matches in fraud scoring Use social presence only as a low-weight corroborating signal, and prevent it from overriding device, behavioural, or transactional evidence when approving accounts or payments.
- Correlate first-party behaviour across sessions Build decisioning around consistent device history, location patterns, and transaction cadence so a single fabricated profile cannot carry the identity through verification.
- Flag identities with compressed backstory formation Review cases where a new email, social profile, and purchase activity appear within the same short window, because that convergence often indicates synthetic assembly rather than genuine customer history.
What's in the full article
Sift's full article covers the operational detail this post intentionally leaves for the source:
- How Sift uses cross-network behavioural history in decisioning workflows
- The way its Sift Score updates in real time as new signals arrive
- Examples of how fraud teams can distinguish a trustworthy returning user from a synthetic identity
- The article's practical discussion of how social data should be weighted inside fraud models
👉 Read Sift's analysis of why social media signals do not equal consumer trust →
Social proof in fraud scoring: what teams are getting wrong?
Explore further
14/09/2026 10:10 am
Social presence is not an identity control, it is a weak corroborating signal. Fraud teams often overvalue visible artefacts because they are easy to inspect and easy to explain, but that makes them vulnerable to manipulation. The better control question is whether the identity shows durable, cross-context behaviour that is difficult to fake at scale. Practitioners should treat profile visibility as supporting evidence, not as proof of trust.
A question worth separating out:
Q: How should security teams respond when synthetic identities pass verification checks?
A: They should treat the pass event as the start of a governance review, not proof of legitimacy. The next step is to examine what evidence was reused, whether the identity can be reused elsewhere, and whether downstream privileges were granted on the basis of a single check. Verification success should not equal broad trust.
👉 Read our full editorial: Social media signals are weak proof of consumer trust
