The star witness in the US government’s landmark trial against Meta believes that a multibillion-dollar settlement reached on Wednesday does not go far enough to stop the social media giant’s harms to young users.

“The limitations that are in the agreement are the equivalent of saying: ‘Well you can smoke as many cigarettes as you can in two hours a day,’” said whistleblower and ex-Facebook employee Arturo Béjar. “It doesn’t make the cigarettes any safer.”

Dozens of states had sued Meta, alleging that it designed addictive products that led to mental health problems for young people such as disordered eating, depression and suicide. The states and the company reached a wide-ranging settlement agreement on Wednesday that will force Meta to make major changes to its platform, including putting strict limits on how much time teens are able to spend on apps like Instagram and Facebook.

Béjar’s testimony formed a key part of the government’s case. The former employee, who worked directly with Mark Zuckerberg, testified the Meta CEO lied about the company’s commitment to child safety and ignored warnings about harmful content.

Meta rejected Béjar’s criticism that the settlement was too narrowly focused on time limits, pointing to other requirements such as a default setting for teens to see a non-algorithmic feed when they open their accounts and parental controls with an option to revert teens’ feeds to non-personalized content.

Many of the changes that Meta must implement following the agreement involve increased parental oversight of app usage and making pre-existing teen safety settings the default for young users. It also creates a “night mode” and “school mode” that restrict app usage during certain times of the day. Several “social comparison” features such as beauty filters and likes will also be disabled by default.

While state attorneys general touted the settlement as a victory and Meta called on other platforms to institute similar policies, Béjar argues the settlement lacks independent oversight and does not necessarily make Meta’s platforms safer for young people.

Béjar says algorithmic design choices that make Facebook and Instagram addictive and harmful remain unchanged by the agreement and its time limits. He had hoped the agreement would have done more to address how social networks’ algorithms steer young users toward content, as well as instituted independent auditing and quantifiable obligations for reducing harm.

“This is not an all clear to say that the product is safe,” he said. “As with tobacco and opioids, the product continues to be harmful after the settlement. The work needs to continue in order to make sure that it’s safe enough for young people.” Meta’s spokesperson argued the platform already has extensive protections for teens that guard them from seeing harmful content or engaging in conversations with adults they don’t know.

“We have a huge raft of built-in protections,” a Meta spokesperson said. “Béjar’s comments really ignore that and they also ignore the huge value that teens get from social media in all kinds of ways.”

A former Facebook safety engineer, Béjar testified for two days during the trial about his view that Meta had adopted a “don’t ask, don’t tell” policy about the safety of children on its platforms.

Meta did not accept any wrongdoing in its settlement and denied all allegations during the trial. The company has been facing an avalanche of litigation this year related to claims that it knowingly created products that hooked young users on its platforms and disregarded harms.