Ofcom has criticised TikTok and YouTube, saying in a new report their content feeds are "not safe enough" for children.
The findings follow the regulator's call for stronger action on children's online safety, and said Meta, Snap and Roblox had each agreed to stronger anti-grooming measures.
Ofcom added it would share concerns that sites were not effectively enforcing minimum age rules with the government, whose consultation on whether to ban social media for under-16s soon ends.
YouTube said it worked with child safety experts to provide "industry-leading, age-appropriate" experiences for children. TikTok said it was "very disappointing" Ofcom had failed to acknowledge its safety features.
Ofcom's criticism is part of a new report into how five large social media and video platforms responded to its demand for stronger protections for children.
"Notably, TikTok and YouTube failed to commit to any significant changes to reduce harmful content being served to children, maintaining their feeds are already safe for children," it said.
"Our wealth of evidence, published today, suggests they are still not safe enough."
In response, TikTok and YouTube pointed to safety features already in place on their apps - including TikTok stopping direct messaging for under-16s and YouTube's short-form video timer, where parents can set time limits for the scrolling Shorts feed.
Social media consultant and analyst Matt Navarra said the criticism illustrated a shift to seeing online harms as "a product problem".
"The old debate was, 'did the platform remove harmful content quickly enough?' - the new one has shifted towards, 'why did the platform show it to a child in the first place?'" he said.
Ofcom has faced frequent calls to take stronger, speedier action to enforce the UK's online safety rules to protect children online - with some critics saying it has failed to do so quickly enough.
Speaking on the BBC's Today programme on Thursday, Ofcom chief executive Dame Melanie Dawes defended its actions but said she knew the job was "not done" yet.
"We're talking about a twenty-year culture at Silicon Valley of not taking safety seriously; you can't change that overnight," she said.
Asked what the regulator would do if platforms did not comply with its rules, Dame Melanie said Ofcom was "ready to take the toughest enforcement action".
"We will absolutely move into a formal investigation if we need to," she added.
In a separate statement before the report was published, Dawes said Ofcom was also "deeply concerned" companies were still failing to take the necessary action to keep underage children off their platforms.
A survey by the regulator found 84% of children aged eight to 12 were still using at least one major service with a minimum age of 13, as it warned stronger legislation may be needed.
Online safety researcher Prof Victoria Baines said the research was "unsurprising" given the "limited success" found so far in removing accounts belonging to under-16s in Australia following its social media ban.
"It may be that some platforms will have to use more behavioural data - what a user is watching, engaging with, and chatting about - to determine whether they really are above the minimum age," she said.
Ofcom's report highlighted changes made by Snap, Roblox and Meta which focused on reducing grooming risks.
Ofcom said Snap, which owns Snapchat, had agreed to block adult strangers from contacting children by default in the UK, stop encouraging children to add people they do not know, and introduce "highly effective" age checks this summer.
A Snapchat spokesperson said it would roll out these measures while "preserving privacy protections and the ability for our community to stay connected with their real friends and family".
The report said Roblox would let parents switch off direct chat entirely for under-16s, while Meta would hide teens' Instagram connection lists by default and develop AI tools to detect likely sexualised conversations in DMs.
Andy Burrows, chief executive of Molly Rose Foundation, a UK-based online safety charity, welcomed the report, calling big tech platforms "complacent and evasive when it comes to protecting children from preventable harm".
He added: "Ofcom will be judged by how quickly it can reduce exposure to online harm. A stronger regulator must be accompanied by a conditional ban on personalised algorithms that continue to push out a tsunami of harmful content to teens."
Ofcom said the promises must now be implemented quickly and properly, warning it will act if platforms failed to deliver.
A government spokesperson said Ofcom had its “full backing” to ensure firms prevented and removed harmful content from their platforms and feeds.
“Too many children are still being exposed to harm online and that cannot continue,” they said.
“That's why we're considering the full range of options - from age limits and app curfews to an outright ban - and will set out next steps by the summer.”
The government's consultation on whether to ban social media for under-16s is due to close on 26 May, with the government planning to respond in the summer.
On Thursday the Education Committee published its response to the consultation, calling for a ban on social media for under-16s.
It also called for urgent action to curb features it said were deliberately designed to drive excessive screen use among under-18s.
But it said a ban should only be seen as a starting point for online safety.
"The Education Committee's recent report is clear - social media firms cannot be relied upon to self-regulate," the committee's chair, Helen Hayes MP, told the BBC.
"Until the safety of children and young people comes before commercial incentives, they will continue to be exposed to the worst of social media and online harms. We need a total reset."
She added: "Only a statutory ban on social media for under-16s, as well as restrictions on addictive and high-risk features for under-18s, will keep children safe from harm."
Sign up for our Tech Decoded newsletter to follow the world's top tech stories and trends. Outside the UK? Sign up here.
