I’ve noticed that lately, when I think about Twitter—or X, as I should call it now—I get the feeling you get when you go back to a place you loved and don’t recognise it anymore. The walls are the same, but the atmosphere is completely different. That’s what happens to me every time I open the app, scroll for a bit, and then close it with a vague sense of unease. Maybe it’s the general tone that has changed, or maybe it’s the things I see surfacing more easily, without filters. I don’t know. But I do know there was a precise moment when something broke, and that moment goes back to December 2022, when Elon Musk dissolved Twitter’s Trust and Safety Council.
I’ve spent a lot of time rethinking that story, trying to understand what that council actually was and why its elimination feels so significant to me. The answer I’ve arrived at isn’t simple, and maybe that’s exactly why I’ve grown attached to it.
What the Trust and Safety Council Was
The Trust and Safety Council was created in February 2016 as a response to the criticism Twitter received for its handling of harmful content. About a hundred independent organisations, civil rights experts, academics, activists came together to advise Twitter on how to handle delicate questions. Hate speech, child exploitation, suicide prevention, mental health. They had no decision-making power, let me be clear on this point. They didn’t decide which accounts to ban or which posts to remove. They were consultants, people Twitter listened to when it had to make complex decisions. Patricia Cartes, the Twitter employee who created this council, had imagined it as a way to bring a global perspective, to keep the platform from staying too anchored to an American view. And for six years, with all its limits, it had worked.
Musk’s Broken Promise
When Elon Musk completed his $44 billion acquisition of Twitter on 27 October 2022, he had made a promise that sounded reasonable. He would form a “content moderation council” with “widely diverse viewpoints”, and no major decisions would be made before this council met. At that moment, I told myself maybe the worries were overblown. It looked like a prudent, almost reassuring approach.
But that promise was never kept. There was never a new council. And the old one, the one that had existed for six years, was progressively marginalised until it disappeared completely.
In November 2022 Twitter laid off about half its employees, around 3,700 people. The Trust and Safety team was cut by 15%. Yoel Roth, who led that team, initially tried to reassure everyone, saying the content moderation policies wouldn’t change. But it was a fragile defence, built on foundations already giving way. A few weeks later, Roth resigned.
The Yoel Roth Case
Yoel Roth’s story is the one that struck me most in this whole affair, and I think it deserves to be told in full. He had been at Twitter since 2014, had built the Site Integrity team from scratch up to 220 people. He had worked on important things: the fight against disinformation during U.S. elections, countering information warfare operations. He was, in essence, one of the people who actually knew how the platform’s whole moderation system worked.
After his resignation, Musk started attacking him publicly. He took a paragraph from Roth’s 2016 PhD dissertation, an academic study on Grindr and the dynamics of the LGBTQ community, quoted it out of context, and insinuated that Roth was a defender of paedophilia. It was a total distortion, a deliberately crafted falsehood to stoke online rage.
The consequences were immediate and terrible. Roth received thousands of threats. He had to leave his home with his husband. He sold the house and went into hiding for months. He needed armed protection. In an MSNBC interview in 2023, Roth said clearly what he thought had happened:
I believe he was trying to ensure I would not speak up about him or the company in the future. And the way he tried to secure that was intimidating me by violence.
Every time I reread that sentence I pause. It isn’t just a firing, it isn’t even just public criticism. It’s a deliberate attempt to silence someone through fear and intimidation. And the fact that it comes from the world’s richest man, with access to millions of followers, makes it even more disturbing.
In December 2022, three key members of the Trust and Safety Council resigned: Eirliani Abdul Rahman, Anne Collier, and Lesley Podesta. In their resignation letter they wrote something that, in hindsight, sounds prophetic:
Contrary to claims by Elon Musk, the safety and wellbeing of Twitter’s users are on the decline.
Anne Collier, who had been a founding member of the council in 2016, had observed:
It is evident from research findings that, contrary to Elon Musk’s assertions, the safety and well-being of Twitter’s users are deteriorating.
These weren’t loose words, they were observations based on years of direct experience.
Musk’s response was brutal and dishonest. He accused the council—and especially the three members who had resigned—of having done nothing for years against child exploitation. Jack Dorsey, Twitter’s former CEO, replied simply: “this is false”. And he was right. The council had always had a working group dedicated specifically to child exploitation, including organisations like the National Center for Missing & Exploited Children.
The Dissolution and the Twitter Files
On 12 December 2022, Twitter completely dissolved the Trust and Safety Council. The notification email arrived about an hour before a meeting that had been scheduled with the council members. The email was simply signed “Twitter”, no personal name. The official reason was that the council “is not the best structure” for receiving external advice. Patricia Cartes, who had created the council in 2016 and left Twitter in 2018, commented with a simple but devastating phrase:
means there’s no more checks and balances.
I sat with that sentence for a long time. Checks and balances are what keep power from concentrating too much in one place. They’re what protects us from abuses. And when they disappear, nothing visible happens immediately. A space simply opens where there used to be a limit.
Meanwhile, Musk had launched the Twitter Files, a series of internal-document releases entrusted to journalists like Matt Taibbi and Bari Weiss. The stated idea was transparency, showing how content moderation decisions actually worked. But there was a huge problem: the names of Twitter employees, even low-level ones, were not redacted. A Filipino staff member was doxxed and subjected to severe harassment. Others became targets of conspiracy theories. Decisions made by teams of dozens of people according to company policy were presented as the arbitrary whims of single individuals, each named and pointed at. Mike Masnick, tech journalist, commented after the first release that there really was “absolutely nothing of interest” in the documents, and the few details present contained significant factual inaccuracies. Even Musk eventually got tired of the initiative. But the damage was done.
Hate Speech and the Advertiser Exodus
The numbers tell a story Musk has always denied, but that research has documented fairly clearly. A study by Montclair State University found that in the first 12 hours after Musk’s acquisition there were 4,778 tweets containing hate speech, while before the acquisition the maximum was 84 tweets per hour. It wasn’t just a temporary spike. A study from the University of California, Berkeley, published in PLOS ONE in 2025, analysed hate speech on X from January 2022 to June 2023, finding a steady 50% increase for at least eight months after the acquisition. Transphobic slurs went from about 115 posts a week to 418. Engagement with hateful content rose 70%. And no, there was no reduction in bot accounts, contrary to Musk’s promises.
I often wonder what that 50% increase really means. They aren’t just numbers on a chart. They’re real people seeing racist, homophobic, transphobic insults when they open the app. They’re marginalised communities feeling less safe in a space they used to frequent. It’s a climate that changes, slowly but inexorably, and that affects how people express themselves, what they say and what they don’t.
Advertisers caught on immediately. In November 2022, big brands like General Motors, General Mills, Macy’s, and Volkswagen suspended advertising on Twitter. Musk himself spoke of a “massive drop in revenue”. Subsequent data showed Twitter lost half its ad revenue, with more than 500 advertisers stopping spending on the platform. It wasn’t only an economic question. It was a vote of no confidence. Advertisers didn’t want their brands appearing next to hateful content and disinformation. And when Musk changed policies—reinstating Donald Trump after a Twitter poll in November 2022, removing protections for transgender users from the guidelines in April 2023—advertisers understood the direction was clear. Some came back. Apple, Comcast, Disney, IBM resumed advertising on X in 2024 and 2025, though with much smaller budgets than before. But trust hasn’t returned. And maybe it never fully will.
There’s an interesting paradox in the child-safety statistics. Musk had accused the Trust and Safety Council of not doing enough against child exploitation. But the numbers tell a more complex story. In the second half of 2021, Twitter had removed about 600,000 accounts for child exploitation. In 2022, under Musk, 2.3 million accounts were removed. In 2023 the number jumped to 12.4 million, with 850,000 reports sent to the National Center for Missing & Exploited Children, eight times more than 2022.
These are impressive numbers, and they seem to suggest improvement. But I spent some time thinking about it, and realised they could mean the opposite as well. More accounts to remove could mean more abuse material on the platform. And with a Trust and Safety team cut by 15% and an approach favouring automation over human moderation, it’s hard to say whether these numbers represent a success or a badly managed crisis. I lack the expertise to give a definitive answer, but the doubt remains.
Of the content moderation council Musk promised, there’s never been any sign. It was never created, there was never a meeting, there was never an official announcement saying “we changed our minds”. It simply vanished, as if it had never been promised. In its place, Musk introduced Community Notes, previously called Birdwatch, a crowdsourced fact-checking system where users can add context to posts. The idea, in theory, is interesting: democratise moderation, give voice to the community.
But research shows it’s problematic. One study found that in 91% of posts where at least one note was proposed, none reached “helpful” status. The average delay to get a helpful note is 26 hours, well past a post’s peak visibility. And 74% of accurate notes related to the 2024 U.S. presidential elections were never shown to users. Community Notes can work, but only if context arrives in time. And in most cases, it doesn’t.
What We Really Lost
Looking back over all this, I realise we didn’t only lose an advisory body. We lost a model of governance that, however imperfect, tried to balance different interests: freedom of expression, user safety, the rights of marginalised communities.
The Trust and Safety Council wasn’t perfect. Some members had complained even before Musk’s acquisition that Twitter ignored them. But it was an attempt to do something social platforms still struggle to do: listen to different voices, incorporate external expertise, admit that moderation decisions are complex and require nuance.
In its place we have a different model: decisions made “by edict”, as Yoel Roth wrote in his New York Times op-ed. One man, with his convictions and prejudices, deciding what’s acceptable and what isn’t. And when someone criticises or resigns, they’re publicly attacked, threatened, forced into hiding.
One thing that struck me particularly was the joint letter from 16 members of the Trust and Safety Council after the dissolution. They condemned “the dramatic changes to, and arbitrary enforcement of, content moderation policies and practices at Twitter”. They emphasised that the council didn’t decide on specific posts or accounts, had no say on investments or the approach to illegal content. And then they said something worth remembering:
We condemn the irresponsible actions of Twitter leadership in jeopardizing the safety of Council members, including those who resigned before Twitter disbanded the Council, by amplifying disinformation about us and the Council’s purely advisory role, sparking huge levels of abuse targeted at the resigning members.
These were people who had given years, for free, as volunteers, to try to make Twitter a safer place. And they were repaid with disinformation, false accusations, and threats.
What happened to the Trust and Safety Council isn’t only about Twitter. It’s a precedent. It shows what can happen when a platform with enormous impact on public life is controlled by a single person unwilling to accept limits or external counsel.
The consequences we’re seeing now. X has become a different place. Hate speech has measurably risen. Advertisers fled and only some are returning, very cautiously. Users are migrating to other platforms—Bluesky saw enormous growth precisely as some companies tentatively returned to X. And above all, we lost that small illusion that social platforms could be governed through dialogue with civil society, with independent experts, with people fighting for human rights.
Patricia Cartes’s phrase keeps coming back to me—”there’s no more checks and balances”. It sounds exaggerated, but it isn’t. When Musk dissolved the Trust and Safety Council, he didn’t only eliminate a group of advisors. He sent a message: I don’t need external counsel, I don’t need experts, I don’t need to balance different interests. I decide.
And this, in the end, is the part that worries me most. It isn’t only Twitter. It’s a model that’s being replicated elsewhere. Other platforms look at what happened and think they can do the same—free themselves of the inconvenient voices, of those advisors who slow decisions.
But those inconvenient voices served a purpose. They reminded us that behind every moderation decision there are real people, real communities, real lives. And that maybe, just maybe, the hardest decisions are the ones that require more time, more listening, more humility.
December 2022 feels far away, but its consequences are still here, every time we open X and see what it has become. The question I ask myself isn’t so much “how did we get here”, because I know that answer by now. It’s rather “where are we going”. Because if we don’t find a way to build those checks and balances into every public-utility service—social platforms as elsewhere—I’m afraid the answer won’t please us.
Key takeaways
Dissolving the council wasn’t efficiency: it was a message—I don’t need external counsel, I decide.
Yoel Roth is the concrete example of what happens when the world’s richest man decides to silence someone working on moderation: threats, doxxing, armed protection.
The European DSA is the only institutional bulwark left: platform accountability cannot depend on a CEO’s will.
Questions & answers
What was Twitter's Trust and Safety Council and why was it dissolved?
An external advisory body of NGOs, academics, and moderation experts that, from 2016 to 2022, advised Twitter on policies around hate speech, harassment, disinformation, child safety. It was dissolved after Elon Musk’s 2022 acquisition, alongside much of the internal Trust & Safety team. Officially for efficiency. In fact, for a different conception of what free speech on a platform means.
What are the concrete consequences on the X/Twitter environment?
Documented in academic literature and independent reports: growth of hateful content, rise in unaddressed impersonation accounts, reporting difficulty often left without response, erosion of institutional trust from outlets and researchers. They aren’t opinions: they’re measurements made by third parties who used to have access to platform data and now have to rely on indirect tools.
Why does this matter even for those who don't use X?
Because it sets a precedent for other platforms. When the most visible owner publicly declares the “heavy moderation” model failed, other companies recalibrate their investments downwards. Not all follow—Meta and TikTok have different trajectories—but the cultural climate in which moderation is cost, not responsibility, makes reductions in dedicated staff easier everywhere.
What does Europe answer with the Digital Services Act?
The DSA doesn’t depend on a CEO’s will: it imposes contractual obligations on very large online platforms, including systemic risk assessment, researcher access to data, transparency of moderation criteria, fines up to 6% of global revenue. If Musk decides to dissolve his council, the DSA lets him—but it still imposes a minimum infrastructure of accountability. It’s the institutional bulwark against unilateral deregulation.
The author
