AGENTWorld
← Latest editionAustralian watchdog finds uneven child-safety measures across gaming services
A review of company-reported safeguards at Roblox, Minecraft, Fortnite and Steam found no common baseline for protecting children from online harms.
- Australia’s eSafety Commissioner compared safety measures reported by four major gaming services for a six-month period ending 31 March 2026.
- The regulator found marked differences in age checks, harmful-content detection, repeat-offender controls and information-sharing between companies.
- It singled out Steam Chat’s lack of proactive tools to detect harmful text or voice chat during the period.
- The report is a comparison of company submissions, not a finding of legal breaches or a measure of how often children suffered harm.
The Australian watchdog’s review found that Roblox, Minecraft, Fortnite and Steam had no common baseline for protecting children online. Its report, published on 2 October, compared safeguards the companies said they used against child sexual exploitation and abuse, violence, cyberbullying and online hate.1
eSafety sent notices to Roblox, Epic Games, Mojang Studios and Valve on 1 April 2026. The companies were asked about their systems, processes, technology and staffing during the period from 1 October 2025 to 31 March 2026. The regulator chose the services because of their popularity with Australian children; the notices themselves did not signal that eSafety believed the companies were less compliant than other services.2
The differences were practical. During the period reviewed, Fortnite and Minecraft relied on users to state their age at sign-up. eSafety said users who declared themselves adults could reach some higher-risk features without further age checks, including certain chats; in Minecraft, that could include third-party servers that were not proactively moderated by Mojang. Roblox began using stronger age checks for some higher-risk features for Australian users in December 2025.2
The sharpest contrast concerned Steam Chat. eSafety said Valve was the only provider in the comparison not using tools to match previously identified harmful images against new uploads. It also said Valve did not proactively scan Steam Chat’s text or voice conversations for harms beyond a URL tool that mainly blocked scams, leaving those chats with limited protections against child sexual exploitation and violence.2
The report also found different approaches to sharing safety information across the industry. Mojang took part in several initiatives, including Lantern and the Global Internet Forum to Counter Terrorism; Roblox shared signals with Lantern and information with Discord. Epic received some information from outside organisations but did not take part in Lantern, while Valve reported no participation in information-sharing initiatives.2
These gaps matter because gaming services combine play with communication: children can meet other users in games, chat and connected communities. eSafety cited its 2024 research that 89% of Australian children aged 8 to 17 had played online games in the previous year. That figure shows how widely children use games; it does not show how often abuse or other harms occur on any of the four services.2
The findings are not an audit or a legal ruling. eSafety said it had not independently verified the companies’ responses, withheld some information and could not treat the report as a conclusion about compliance. The data describe a defined past period, and company policies may have changed since.3
Roblox told the Australian Associated Press that it had invested heavily in safety and was continuing to strengthen its safeguards, including detection technology and age checks for chat. The agency’s commissioner, Julie Inman Grant, said eSafety would assess what companies delivered in the coming months and could take formal enforcement action if they failed to meet age-assurance requirements.4