CHAPTERS
- 0:00 – 1:00
AI’s Pandora’s box: why trust & safety must keep up with innovation
Glen frames AI as an irreversible shift: kids and everyday users are now interacting with systems that can be exploited at scale. He argues that without effective trust and safety, mounting harms will trigger regulation that slows overall innovation.
- •AI is already embedded in daily life, including for children
- •Harms range from spam/“AI slop” to contextual phishing and non-consensual imagery
- •Unchecked abuse invites regulation that can slow product innovation
- •Need proactive systems that prevent harm while enabling faster deployment
- 1:00 – 1:32
What Cinder is: a “club bouncer” platform for enforcing online rules
Glen introduces Cinder as a centralized trust & safety operations platform that helps internet businesses define and enforce policies across users. He positions the company as infrastructure for moderation and abuse prevention, backed by major investors.
- •Cinder centralizes trust & safety operations for online platforms
- •Customers set policies; Cinder helps enforce them consistently
- •“Club bouncer for the internet” metaphor clarifies the product role
- •Company raised $14M from Y Combinator, Accel and others
- •Mission focus: making the internet safer
- 1:32 – 2:02
9/11 upbringing to national security: the origin of a mission-driven career
Growing up in New Jersey during 9/11 shaped Glen’s desire to contribute to a larger mission. That motivation led him toward roles in U.S. government focused on preventing major security threats.
- •9/11 was a formative event influencing career direction
- •Desire to work on a “greater mission” to protect society
- •Interest in agencies like NSA/CIA/FBI as a path to impact
- •Early focus on preventing severe real-world harm
- 2:02 – 2:33
From counterterrorism to Meta: threat intelligence and the red-team mindset
Glen describes moving from government counterterrorism work into Facebook/Meta’s threat intelligence efforts, which invested early before harms became widespread. He highlights the value of red-teaming—thinking like adversaries—to design resilient defenses.
- •Transition from government work to Facebook threat intelligence
- •Meta invested in threat intel before harms fully manifested
- •Red-team work: “try to break things” to find weaknesses
- •Adversarial mindset is essential because attackers adapt
- •This mindset becomes part of Cinder’s team culture
- 2:33 – 3:03
Spotting a widespread industry gap: companies drowning in platform abuse
Through conversations with other companies, Glen saw many teams struggling to understand and control abuse on their platforms. The persistent pull of the problem overcame his initial hesitation about being “ready” to found a company.
- •Widespread struggle across companies to grasp abuse on-platform
- •Problem became an obsession outside of work
- •Founder hesitation: feeling underqualified or early-career
- •Initial plan: join a smaller company and build tools internally
- 3:03 – 3:33
The “build it once” insight: centralizing trust & safety infrastructure
After exploring multiple places to build internal solutions, Glen realized the same core system was needed across many companies. That realization pushed him to found Cinder as a shared, centralized platform rather than a one-off internal tool.
- •Many companies needed similar trust & safety capabilities
- •Centralizing infrastructure is more efficient than repeating builds
- •Founding decision driven by clear business logic and urgency
- •Goal: stop the “underbelly” harms so others can innovate quickly
- 3:33 – 4:03
Getting into Y Combinator: last-minute application and the cofounder requirement
Glen recounts applying to YC shortly before the deadline and getting an encouraging call from Michael Seibel. The key condition: he needed to recruit a cofounder, which forced an early push to assemble a credible founding team.
- •Applied to YC 1–2 hours before the deadline expecting nothing
- •Michael Seibel expressed strong interest in the idea
- •YC requirement: find a cofounder to move forward
- •Challenge: convincing strong talent to leave Facebook/Meta
- 4:03 – 4:36
Assembling the founding team: recruiting trusted ex-Meta collaborators
Within a few months, Glen built a team of people he deeply respected and wanted to work with. He emphasizes that team formation was a defining moment and remains central to execution and company culture.
- •Successfully recruited a strong early team within ~3 months
- •Many teammates came from prior Facebook/Meta collaboration
- •Team quality and trust were foundational to momentum
- •Personal significance of building and working with the team daily
- 4:36 – 5:06
The scale of modern harm: from nation-state capabilities to teenagers with laptops
Glen illustrates the magnitude of online abuse by citing over 100,000 reports to NCMEC submitted by Cinder customers in a single year. He warns that advanced abuse techniques have become democratized, making robust defenses urgent.
- •Cinder customers submitted 100,000+ NCMEC reports in a year
- •This number is “scratching the surface” of broader harms
- •Capabilities once limited to nation-states are now widely accessible
- •Without countermeasures, online spaces risk becoming a “hellscape”
- 5:06 – 6:07
Hard lesson in product direction: the first platform customers didn’t want
Cinder initially built a threat-intel-style platform for trust & safety, based on prior experience and perceived need. Market feedback contradicted their assumptions: customers wanted an operations system, prompting a major product pivot.
- •Building a platform is difficult: integrations, configurability, workflows
- •Initial product vision: threat intel platform for trust & safety
- •Early customer feedback: interesting, but not what they’d buy
- •Pivot toward “overall operations” as the real pain point
- •Core takeaway: tighten around the true customer problem before scaling
- 6:07 – 6:37
Customer empathy as a competitive advantage: solve real workflows fast
Glen stresses that deeply understanding customer pain—how abuse happens and how teams respond—was crucial, even before the product existed. Listening, advising, and quickly returning with a solution helped Cinder build something teams actually needed.
- •Career focus: preventing people from exploiting systems and others
- •Empathy with trust & safety teams shapes product requirements
- •Value even pre-product: listen, diagnose, suggest fixes
- •Rapid iteration: return weeks later with a solution to a real problem
- 6:37 – 8:08
The “Decision Spectrum” and burnout: managing simple flags through complex adversaries
Cinder uses a “Decision Spectrum” framework to categorize moderation work from simple classification to complex investigations of sophisticated networks. Glen highlights a core operational challenge: overwhelming volume, prioritization difficulty, and the risk of burnout for reviewers.
- •Non-consensual intimate imagery (NCAI) is a prominent gen-AI harm
- •Decision Spectrum: simple decisions (nudity/fraud) to complex investigations
- •Volume makes review and prioritization extremely difficult
- •Everything feels like a harm, creating emotional strain and burnout
- •Systems must help triage and scale decisions responsibly
- 8:08 – 9:43
Mission-driven resilience: why purpose and firsthand experience matter
Glen argues that effective trust & safety products require both direct experience with real harms and strong mission alignment because the work is nuanced and emotionally taxing. The team stays motivated by tangible wins—taking down abuse and preventing victimization.
- •Founding team’s firsthand experience informs product credibility
- •Purpose is necessary to sustain work that is difficult and nuanced
- •Team motivation comes from concrete impact stories and takedowns
- •Avoiding despair: there is a path to combat harm, and someone must do it
