An independent game developer for Roblox, someone who builds content on the platform and volunteers for a child safety organization, just told BBC Radio 5 Live something extraordinary:
"When playing Roblox, children need to be monitored 24/7. And if that's not possible, then they shouldn't be playing Roblox."
This is someone who sees, from the inside, what actually happens on the platform. Someone who has witnessed users being lured into inappropriate interactions with strangers. Who has seen people leading children off-platform to have conversations away from the game's safety systems. Who has reported games depicting mass shootings at Sandy Hook and Columbine, and recreations of Epstein Island, and says only about 30% of safety reports get accepted.
Roblox averaged more than 80 million daily players in 2024. Roughly 40% of them, 32 million children, are under 13.
And the best advice an insider can offer parents is: watch them every second, or don't let them play.
That is not a safety strategy. That is a confession that the platform's safety infrastructure is insufficient.
The Problem with "Trust Parents to Make Their Own Decisions"
Roblox CEO Dave Baszucki told the BBC in 2025: "If you're not comfortable, don't let your kids be on Roblox." His Chief Safety Officer, Matt Kaufman, points to mandatory age verification rolled out in January 2026 and advanced safeguards and filters.
These responses share a common flaw: they place the burden of safety on parents and on reactive systems that respond after harm has occurred.
Age verification confirms who a user is. It does not detect what they do after they log in.
Content filters catch known harmful material. They do not catch the groomer who spends three weeks building trust with a child through perfectly innocent-sounding messages before leading them off-platform.
Reporting systems rely on someone, often a child, recognizing danger and taking action. And as the developer told the BBC, only a fraction of reports result in action.
The Australian eSafety Commissioner's Safety by Design framework puts it simply: "The burden of safety should never fall solely upon the user." Telling parents to monitor 24/7 is the opposite of this principle. It places the entire burden on the most overwhelmed and least technically equipped person in the equation.
What 24/7 Monitoring Actually Requires
Let's take the developer's advice seriously for a moment. What would it actually take for a parent to monitor their child 24/7 on Roblox?
They would need to sit beside their child for every minute of gameplay. They would need to read every chat message in real time, across multiple games, often with overlapping conversations. They would need to understand the context of each interaction: is this person a school friend or a stranger? Is this conversation normal banter or the early stages of grooming? They would need to recognize manipulation tactics that trained criminologists study for years to identify.
No parent can do this. Not because they don't care, because it's humanly impossible. A groomer doesn't announce their intentions. The behavioral pattern that leads to exploitation unfolds across dozens of seemingly innocent interactions over days and weeks.
This is not a problem that human monitoring can solve at scale. It is a problem that behavioral detection technology was designed to solve.
Behavioral Detection: What Platforms Should Be Doing Instead
The threats that the Roblox developer described, luring, off-platform contact, grooming, are not content problems. They are behavioral problems. They follow documented patterns that unfold over time:
- Stage 1. Contact initiation: "What game do you play? We should team up!" Every content filter sees this as safe. Because it is, in isolation.
- Stage 2. Trust building: "You can tell me anything. I understand you better than your parents do." Still safe to a keyword filter. But a behavioral detection system recognizes the isolation pattern: separating the child from trusted adults, creating emotional dependency.
- Stage 3. Secrecy: "Don't tell anyone about our conversations." No profanity. No explicit content. No trigger words. But the behavioral sequence, contact → trust → isolation → secrecy, matches documented grooming patterns with high confidence.
- Stage 4. Off-platform migration: "Let's talk on Discord/Snapchat/WhatsApp instead." This is the moment the child leaves the platform's safety systems entirely. The developer specifically flagged this as something he has witnessed and that Roblox does not allow, but cannot effectively prevent.
A behavioral detection system identifies this pattern in stages 1-2, before the child is ever at risk. Not by reading individual messages, but by analyzing how the conversation evolves over time, the same way a criminologist would analyze a case file, but in real time, at scale, across millions of concurrent conversations.
The Technology Exists Today
At Tuteliq, we build exactly this kind of behavioral detection. Our API analyzes conversations for grooming, coercive control, bullying, radicalization, and the rest of our 13 harm categories, in real time, across text, voice, image, and video, in 32 languages (English generally available, 31 more in beta).
We already have a native Roblox integration. The technology can be embedded directly into platform infrastructure, running silently in the background, detecting behavioral patterns that no human moderator or content filter would catch.
When our system detects a grooming pattern, it doesn't just flag the message. It generates a structured incident report identifying the grooming stage, the evidence, and age-appropriate recommended actions, different guidance for the child, the parent, and the moderator. All in under 400 milliseconds.
This is what Safety by Design looks like in practice. The platform takes proactive responsibility for user safety. Parents are informed when there's a real threat, not expected to be the first line of defense.
What Needs to Change
The Roblox developer's interview should be a wake-up call, not just for Roblox, but for every platform where children interact with strangers.
For platforms: Age verification and content filters are necessary but insufficient. If your best advice to parents is "monitor 24/7," your safety infrastructure has a gap that behavioral detection can fill. The regulatory pressure is real: KOSA, still a bill rather than law, is drafted around protection from a defined list of harms, the EU DSA mandates risk assessment for minors, and the UK Online Safety Act requires proactive detection. Countries are banning platforms outright. Russia, Turkey, and Indonesia have all restricted or banned Roblox. The question is whether platforms will invest in proactive safety, or wait for governments to decide for them.
For parents: You should not need to be a 24/7 surveillance system. Your children deserve platforms that protect them by design, not platforms that outsource safety to you. Ask the platforms your children use: what behavioral detection do you have? How do you detect grooming patterns? What happens when a stranger tries to build an inappropriate relationship with my child? If the answer is "we have content filters and a report button," that is not enough.
For regulators: The developer's testimony confirms what behavioral research has documented for years: the most dangerous threats on platforms are invisible to content classification systems. Regulations that focus only on content removal, while important, miss the behavioral threats that precede exploitation. Safety standards should require behavioral detection capabilities, not just content moderation.
No Child Should Need a 24/7 Bodyguard to Play a Game
The developer who spoke to the BBC did something brave. He told the truth about what he sees on a platform used by tens of millions of children every day.
But "monitor your children 24/7" cannot be the answer. The technology to detect behavioral threats in real time exists. It can be integrated into any platform in days. It works across text, voice, and video. It catches the patterns that content filters miss and human moderators can't scale to cover.
The question is not whether this technology is available. The question is whether platforms will deploy it before the next child is harmed.
Every child deserves to just be a child. Happy and safe. Even on Roblox.
---
Gabriel Sabadin is the CEO of Tuteliq AB, a behavioral detection API for online safety. Dr. Nicola Harding, former criminologist at Lancaster University, contributed to this article.