Significant gaps remain in child-safety measures on Roblox, Minecraft, Fortnite and Steam, according to a new report from Australia's eSafety Commissioner.

The investigation examined what online gaming companies are doing to protect children from sexual exploitation and abuse, cyberbullying, violent content, online hate and extremist influences.

On April 1, 2026, eSafety issued legally enforceable transparency notices to Roblox, Mojang Studios, which operates Minecraft, Epic Games, which operates Fortnite, and Valve, which operates Steam and Steam Chat.

The report, released in October and based on the companies' responses, identifies substantial differences in how they verify users' ages, monitor communication features used by children, and detect and remove harmful content.

eSafety found that Minecraft and Fortnite relied heavily on users declaring their own ages when setting up accounts.

In some circumstances, users who registered as aged 18 or over could access higher-risk communication features without further age checks.

The report also notes that Minecraft users who declared themselves adults could play on third-party servers that Mojang Studios did not directly moderate. eSafety warned that these environments could expose children to unlawful or harmful content.

The report identifies important gaps in Steam and Steam Chat as well.

Valve did not use hash-matching technology to detect previously identified child sexual exploitation imagery. It also did not use tools to proactively detect sexual exploitation or violence-related harms in Steam Chat conversations, apart from a limited URL-detection tool primarily aimed at blocking scams.

eSafety also reported that Valve did not have a dedicated trust and safety team for Steam and Steam Chat, relying largely on content moderators and engineers.

Some shortcomings were identified at Roblox. Its Image Auto-mod tool, used to detect violent or child sexual exploitation imagery, had a recall rate of about 51 per cent. This measures the proportion of harmful images in the evaluated material that the tool detected; it is not a measure of overall accuracy.

However, eSafety also acknowledged that Roblox had strengthened its age-checking arrangements. From December 3, 2025, the company introduced stricter age assurance in Australia for chat and certain other higher-risk features.

The report also recognised some of Epic Games' safety measures for Fortnite. Epic Games said every new user-created gaming island was reviewed before being made available to players.

At the same time, eSafety reported a median review time of 18 hours and 53 minutes for certain material flagged by Epic Games' automated tools as potentially involving child sexual exploitation. The report makes clear that the flagged material was ultimately found not to violate the rules in each case. Epic Games said the material was unavailable to users while it was being reviewed.

Minecraft also took longer than other providers to respond to some user complaints about harmful content, according to the report.

Australian children's widespread use of online games provides an important backdrop to the investigation.

Earlier eSafety research found that 89 per cent of Australian children aged 8 to 17 had played online games during the previous year.

Among young online gamers, 79 per cent had played with others, while 40 per cent had played with people they did not know offline.

eSafety Commissioner Julie Inman Grant has warned that, while online games offer children entertainment, learning and social connection, some gaming platforms can also be used by offenders to make initial contact with children.

eSafety has previously warned about offenders establishing contact in gaming environments and then moving children to private messaging services, where sexual exploitation and other harms may occur.

The regulator is urging gaming companies to strengthen safeguards, including reliable age assurance, controls on contact with strangers, proactive detection of harmful images and conversations, and measures to prevent banned users from returning through new accounts.