At a glance
WHAT IT’S REALLY ABOUT
Cinder’s founder on scaling AI trust and safety operations platform
- Wise argues that AI is accelerating online harms—from spam and contextual phishing to non-consensual intimate imagery—creating urgent demand for stronger trust and safety systems.
- Cinder positions itself as a configurable “club bouncer for the internet,” enabling platforms to set and enforce moderation rules across users through one centralized operations layer.
- His career path (post‑9/11 motivation, government counterterrorism, Facebook threat intelligence/red teaming) shaped Cinder’s adversarial mindset and product direction.
- A key startup lesson was a major pivot: customers didn’t want a standalone threat-intel tool; they needed end-to-end trust and safety operations, forcing the team to refocus on the core workflow problem.
- Wise emphasizes the human and organizational cost of moderation (triage overload and burnout) and frames mission-driven teams and better prioritization frameworks as essential to sustainability and impact.
IDEAS WORTH REMEMBERING
5 ideasAI safety failures will trigger regulation that slows innovation.
Wise’s central argument is that if platforms don’t proactively stop severe abuses, governments will intervene, forcing pauses and compliance-heavy processes that reduce the speed of product deployment.
Centralized trust and safety infrastructure can be a force-multiplier.
Instead of each company building bespoke tooling, Cinder aims to provide a shared enforcement and workflow system so many platforms can implement policy rules and operations consistently.
Customer reality beats founder intuition—platforms must solve the operational workflow first.
Cinder initially built a trust-and-safety “threat intel platform,” but customers signaled they needed a full operations solution (case management, workflows, configurable data/signals), prompting a strategic pivot.
Building a platform product is slow because configurability is the product.
Wise highlights that a true platform must ingest many signals, support multiple content types, and map to each customer’s unique data models and review processes—work that is easy to underestimate.
A spectrum-based triage model helps match tools to harm complexity.
The “Decision Spectrum” frames problems from simple classifications (e.g., nudity/fraud detection) to complex adversaries and networks, implying different automation, investigation, and prioritization approaches.
WORDS WORTH SAVING
5 quotesThe Pandora box of technology is already open.
— Glen Wise
Someone needs to be able to stop those horrible things from happening so innovation can continue.
— Glen Wise
You can think of Cinder as the club bouncer for the internet.
— Glen Wise
We see abuse today that four years ago only a nation-state could do, and today a teenager with a laptop is able to do it.
— Glen Wise
As a society, we need systems that can combat this abuse before it turns into a total hellscape.
— Glen Wise
High quality AI-generated summary created from speaker-labeled transcript.
