How Feeld Secured Authentic Connections by Scaling Moderator Expertise

As one of the fastest growing dating apps in the world, Feeld’s success is built on a core promise: empowering people to pursue their desires in a safer and open environment. While that vision drew mostly legitimate members looking for something different in the dating space, the brightest lights cast the longest shadows and bad actors took note as well. It wasn’t long before these scaled networks were attempting a familiar playbook – mass sign-ups, accounts lying dormant to evade detection, and sudden and orchestrated re-activation to coordinate spam schemes.

Building on an already high standard of safety, Feeld’s expert moderation and engineering teams set out to raise the bar even higher. The team focused on anticipating new threats and leveraging their moderators' expertise to create an adaptive defense system capable of stopping new attacks before they could impact members.

“A consent-driven experience is at the heart of a healthy community on Feeld. Like most dating platforms, we face ongoing challenges from fraud, scammers, and bots — but our approach is grounded in awareness and adaptability. By evolving with the landscape of online safety, we’re better equipped to keep Feeld a place where members can connect and express themselves with confidence.”
- Feeld Trust & Safety leadership

From Overwhelmed to Empowered

Instead of simply adding more people to the moderation queue, Feeld partnered with Musubi to fundamentally change their defense strategy. The solution was AiMod, an agentic AI system designed to learn directly from Feeld’s human moderation team.

This created a powerful, human-in-the-loop defense system:

  • Strategic Focus: By deploying AiMod to handle the high volume of reported members, the moderation team was free to  focus on strategic tasks: identifying new, emerging threat patterns, and making thoughtful high-quality decisions to protect members.
  • Real-Time Learning: Every decision the human team made became a powerful signal for AiMod. The AI analyzed these expert actions to continuously train itself on Feeld’s unique and evolving moderation standards.
  • Intelligent Automation: With this training, AiMod began to automate the enforcement of high-confidence cases at a massive scale, eliminating bot networks almost instantly with extreme precision and virtually no false positives.

An Adaptive Shield Against Evolving Threats

The true power of this approach was its adaptability. As moderators identified bad actors’ new tactics, their decisions fed into AiMod, creating a custom clustering graph across a diverse set of signals. AiMod learned to see a bot not only by its message content, but also by its user details and behavioral trends. These included device hopping, high-risk IP addresses, unusual sign-up patterns, and periods of dormancy followed by sudden, high-volume activity.

Each day, as bad actors adapted, AiMod adapted with them, keeping the platform safe and the spam issue at bay. This approach yielded results that were not only faster but also more accurate than human moderation alone:

  • Dramatically Faster Decisions: Musubi's AiMod makes decisions up to 500x faster than human moderators, shrinking the window for bad actors to operate from hours to seconds.
  • Higher Accuracy: Across over 100 million real-world decisions, AiMod has proven to make 89% fewer mistakes than traditional moderation.
  • More Effective Enforcement: AiMod catches 64% more bad actors, proactively identifying threats that might otherwise be missed.
“It’s critical that we maintain a low false-positive rate when enforcing platform bans — we want to keep bad actors out without unfairly blocking genuine members from connecting on Feeld. We like to think of it as keeping the dolphins safe from the tuna nets. Musubi has been an excellent partner in this; AiMod is nuanced enough to know when to defer to a human, yet decisive enough to act instantly when the call is clear.”
- Feeld Trust & Safety team

Fostering Authentic Connections

Musubi has transformed Feeld’s moderation cycle. Their human moderation team now operates as the strategic intelligence that powers an automated, at-scale defense system. This human-AI partnership ensures Feeld can continue to grow rapidly while staying true to its core mission of providing a safer, bot-free environment where authentic connections can thrive.