Trump Administration's Arrival on Bluesky Highlights Growing Pains for Open Networks

Last week, the United States Department of Homeland Security—last seen brutalizing journalists, children, and clergy in the streets of Chicago—joined the Bluesky social internet platform and began, with the White House and other federal agency accounts, trolling and threatening Trump administration political opponents.

It’s no accident that US political operatives have been singling out Bluesky as a target for political control and retribution for months. The containment of mass-scale social media by the authoritarian-friendly billionaire class is just about complete: TikTok is coming under the control of billionaire friends of the Trump administration, X is led by the the same billionaire who took a wrecking ball to vital government services, and Meta is run by another billionaire with a miserable record on human rights and misinformation. Open social networks—like those running on Bluesky’s AT Protocol or the Activity Pub protocol that powers Mastodon—are the only remaining leak in their information-containment fields.

With about 39 million account-holders, Bluesky is still tiny compared to the mega-platforms, but by the numbers, it’s the most successful open social network of the moment. Thanks to its relative ease of use and commitment to the basic moderation work mass-scale platforms increasingly decline to perform, Bluesky has also become the alternative network of choice for academic, cultural, and left-leaning journalism communities who have left X or are simply hedging their bets.

The administration’s arrival on Bluesky comes after several months of centrist commentary conflating the network with a Twitter-native brand of scoldy leftism in which internet repliers are rude to centrist and right-wing pundits—a predictable follow-on to 2024 coverage of Bluesky as a “Blue Heaven.” Right-wing trolls and commentators have won national media attention for having transphobic posts labeled on Bluesky or being suspended for spamming replies with political memes. More recently, right-wing media exaggerated Bluesky’s role as a haven for threats of political violence and recommended that the Justice Department “extensively investigate” the platform (and that other notorious hotbed of leftism, Reddit) for possible commission of unspecified crimes.

Astute network observer Laurens Hof notes that the prevailing centrist commentary constructs Bluesky both as an easily dismissed minor player and an entirely partisan force. The right-wing demonization of the platform stacks atop this framework and takes the next obvious step: if Bluesky is an organizing base for an opposition the right wing characterizes as “leftist terrorism,” the platform is a threat to public safety and should be suppressed.

Against this background, the administration’s antagonistic entry to the platform is best understood as a game of chicken: If Trump administration accounts unambiguously break Bluesky’s terms of service, as they seem almost certain to do, the company will have to delete the posts or suspend the accounts and face the administration’s retaliation, alter its terms of service to accommodate Trumpist trolling, or ignore the breaches. None of these options are good for the company, which will have to decide how much it wants to be the focus of an authoritarian power’s noisy tantrums, how much time and money it can afford to spend defending itself against a regime that openly conducts political prosecutions, and how far it can push a user base already uneasy after months of hotly contested moderation controversies.

And for the AT Protocol ecosystem (affectionately, the atmosphere) as a whole, the stakes are high: The company is already moving governance of the protocol itself to the Internet Engineering Task Force (IETF) and has issued a patent non-aggression pledge in service of its initial stated aims; these measures may be protective for the atmosphere, but it’s not clear that the ecosystem is stable enough—or decentralized enough—to withstand the loss of Bluesky the company.

Meanwhile, the 39 million account-holders on Bluesky inhabit a network whose layers of technical and cultural shelteredness have sharply eroded since the network launched in 2023: the initial login wall has long been discarded and the network’s early obscurity has failed to make it a less-attractive target for authoritarian state actors and harassment networks. Beyond that, simmering tensions between the company and several early-adopter communities have reduced trust in the company’s willingness or ability to take unpopular stances to protect groups under attack by culture warriors in the federal government. Although Bluesky has held the line on many basic anti-harassment and anti-bigotry moderation policies, neither the open, public framework at the heart of the company’s mission nor its structural and policy preference for distributed moderation is meeting the needs of many of its most vulnerable users, who want—and loudly call for—features and policies the company is unwilling or unable to provide.

So how did so many people seeking shelter from (mostly) right-wing harassment and state repression end up on a locked-open network devoted to big-world networking and devolved governance of speech?

How we got here

There’s no evidence that the majority of Bluesky’s users came to the platform because of a sustained interest in decentralized networking or distributed moderation. Many may have been drawn—to Bluesky or to the fediverse—by the possibility of a billionaire-proof network that couldn’t be captured, enshittified, or turned into a political tool. Most, though, seem to have come to Bluesky because it was the easiest and most pragmatic option for people who could no longer tolerate X.

A brief chronology makes it clear how much events at Twitter/X—and to some degree, on larger social platforms—have driven the migration of US-based accounts, first to the fediverse and then, in greater numbers, to Bluesky:

• In February of 2022, Bluesky hired its first three employees. Eight months later, Elon Musk took ownership of Twitter and began dismantling the platform’s moderation and anti-misinformation policies and practices. A wave of account migrations from Twitter into the fediverse (mostly Mastodon) began.
• In February of 2023, Bluesky began the closed beta for its reference app, a Twitter-like prototype also called Bluesky. The next month, Twitter—by then X—fired more content moderators, and Meta and YouTube, too, spent the spring of 2023 recanting their previous commitments to healthy discourse and the fight against misinformation. A wave of migrations from X to the still-invite-only Bluesky app began. Meta’s closed (but potentially open in the future) Twitter competitor, Threads, also opened that summer and reported an immediate surge in new accounts, though its tight integration with Instagram has made its initial and subsequent usage numbers hard to interpret.
• In February of 2024, Bluesky opened account registration and allowed non-logged-in users to view most accounts and posts. Throughout 2024, Elon Musk used his account and his role as platform executive to campaign for Trump’s presidential candidacy. Brazil’s judicial ban of X in September of 2024 brought a wave of Brazilian users to Bluesky, and after Trump’s reelection in November of that year, a wave of Americans left X—and many of them landed on Bluesky. Subsequent malfunctions and moderation changes at X continued to push accounts toward open networks.

The most obvious reason for migrations to open networks is that people wanted quite badly to get off of X without losing touch with their communities. Mastodon, which has had higher friction on joining and re-connecting broken ties across communities, was hard for a lot of people. Community migration to Bluesky was easier, both initially in 2023 and then much more so in 2024, when Starter Packs launched and made one-click cross-network community recovery much closer to reality.

As a result of close connections between changes at Twitter and the growth of Bluesky’s user base, there’s a mismatch between what Bluesky appeared to many people to be—essentially, Twitter minus Elon Musk—and what the platform actually is and does. The problem, for many users, is twofold: First, the platform’s design affordances look the same as those on familiar closed networks, but it’s actually something else entirely. Second, the company’s commitment to doing moderation at all seems to have led many people to believe that it would do high-context, firmly progressive, community-style centralized moderation of the growing global platform, which it isn’t.

Looks like a duck, walks like a yeti

On the design affordances side, Bluesky looks like Twitter/X, but it works differently because it doesn’t have the same goals. Bluesky makes available many familiar features: you can follow, interact, block, mute, upload media, add alt text to images, and use lists much as you could on Twitter. But you can’t (yet) make private accounts, because the protocol is built for open, public networking; you can’t switch a post or account from public to private for the same reason.

Bluesky also offers things you can’t find on most other services: Distributed moderation makes it easy to label, mute, or block accounts based on a trusted source’s recommendations and taxonomies. Fine-grained interaction controls allow Bluesky users to tap the brakes on unwanted interaction cascades. Custom feeds are easy to make and can surface posts from a given community, topic, or link type (like Paper Skygest for academic publications and Gift Links for sharing gaps in media paywalls). The protocol’s architecture makes it easy to build services that, for example, allow users to mass-block or mute all followers of a troubling account, or everyone who liked a post they find repugnant. And these innovative features come with new and sometimes ill-understood risks or downsides.

Ultimately, Bluesky offers a different constellation of features and options than either the closed platforms or earlier decentralized ecosystems like the fediverse. And—maybe particularly in moments of widespread societal stress—new options aren’t what everyone wants. But I think they might be what we’re going to need. Which brings us to the second conflict between Bluesky the company and several communities in its early user base: the gap in expectations of its centralized moderation.

Platform moderation vs. community management

Despite its commitment to moderating identity-based violence and hate campaigns, Bluesky is not offering the kind of high-context, community-led centralized speech governance that many users want. Rather than offering a tour of every controversy since the company’s inception, I’ll point to just two of the longest-running: The continued presence of Jesse Singal on the network, and the deletion of fundraising accounts associated with Palestinians in Gaza. In short, the company will moderate anti-trans slurs on the platform, but it won’t ban a journalist known for spreading misinformation and undermining access to gender-affirming healthcare—because, by Bluesky’s accounting, he didn’t break their terms of service and they don’t ban users based on their behavior outside the platform. And despite pushback from users and media attention, Bluesky hasn’t collaborated with organizations who could have helped them enact validation and contextually sensitive moderation for accounts claiming to represent Palestinians in Gaza—or at least, they haven’t given any public signs of such actions.

In the Digital Infrastructure Insights Fund–funded fediverse governance study I undertook with Darius Kazemi in 2023, we used the term “high-context moderation” to describe the kind of governance many users want Bluesky to perform, writing that:

…thoughtfully governed, medium-sized Fediverse servers are especially well positioned to offer a model of high-context, culturally sensitive online community that outperforms most interactions with centralized platform governance.

High-context moderation work integrates fully with specific communities to understand their needs and desires. Depending on community needs, it may take off-platform behavior into consideration and employ power analysis in service of community safety and justice. This is the kind of culturally attuned moderation that the fediverse model can and sometimes does enable. It’s also the kind of moderation that Blacksky Algorithms employs and that Northsky Social is working toward in the ATmosphere—and that the AT Protocol was, in theory, built to support.

This kind of moderation is structurally incompatible with one-size-fits-all global corporate governance, even before state actors begin manipulating and censoring networks to suit their own ends. Bluesky has made many good moderation decisions—most notably the decision to moderate at all, which immediately lost it the support of its early champion, Twitter co-founder and former CEO Jack Dorsey. And it’s also made several unpleasant decisions that nevertheless align with what the company has always claimed to be, and a few chaotic and bad calls. But I don’t think platform companies should—or can—act as the governance layer of our online lives. In this, I share tech journalist and Bluesky board member Mike Masnick’s informed belief that global centralized moderation is a dead end. And if centralized moderation isn’t a plausible way forward, then experimentation toward genuinely community-led governance of decentralized networks is necessary.

My own hope for Bluesky has never been that Bluesky itself would be a durable refuge for vulnerable people, but that it would—alongside Mastodon’s new board and many fledgling initiatives—lay the necessary foundations for self-governance by communities that have the skill and devotion required to take care of their members on the open social internet. People aren’t misguided to want shelter from surveillance and harassment, or to seek high-context, community-centric moderation; all echo-chamber punditry notwithstanding, those are necessary conditions when life online has such direct and dramatic effects on the character of our offline societies.

Bluesky the company is extremely unlikely to provide those things—not directly, not for all communities, and not in the ways many people need. But community-focused network projects built on top of AT Protocol—and Activity Pub, and the open protocols that will emerge in five or ten or fifteen years—remain our best way out of the manipulative societal (and sometimes quite literal) nightmare so much of the social internet has become.

But what about now

It would have been better if the open networking sector had longer to build—just another year or two before Musk’s acquisition of Twitter or before the acceleration of the global authoritarian slide. But the ecosystems we have today offer plentiful ways for people who care about community self-governance to support the work of building out viable systems for high-context governance and connecting them together, by supporting efforts like Blacksky Algorithms and Northsky Social, which are building AT Protocol-based social networking stacks centering Black and 2SLGBTQIA+ communities, Bonfire Networks, which is building next-generation public-interest networking on the Activity Pub protocol, and A New Social, which is building bridges and migration services that connect people and communities across open networks.

For people on Bluesky now, the Trump administration’s arrival should serve as a useful reminder that AT Protocol is built for fully open, fully public networking. With the exception of direct messages and internal records on moderation decisions and company strategy (significant exceptions), Bluesky the company holds very little data that isn’t already public and viewable through third-party services, which means it can’t be forced to turn over messages from private accounts, for example, because those don’t exist. The surveillance risks for Bluesky users are probably not in real terms greater this week than they were before, but it’s unclear that the majority of Bluesky’s users understood these risks to begin with. And the risk of being harassed and threatened by US government accounts and the followers they brought from X to troll Bluesky users has certainly increased.

There’s civil society work to be done helping people in new networks understand the new risks they face as they escape closed and increasingly unlivable megaplatforms. Safety guidance designed for vulnerable users on X or Meta platforms doesn’t translate 1:1 to Bluesky—or to Mastodon and other fediverse platforms. Institutions and communities under attack should publish informed guidance for their members on the risks of open networks, and the best way to use open networks’ existing safety features to reduce vulnerability to attacks by state actors, operatives, and freelance harassers.

On the individual level, people seeking private social networking may be better off, for now, finding a trustworthy Mastodon server and maintaining their connections with accounts on Bluesky via network bridges. On the ATmosphere, people seeking an approach to governance firmly rooted in mutual aid and community self-determination can join Blacksky now and/or financially support the project.

In the meantime, don’t put anything you wouldn’t want handed to the US federal government in your Bluesky (or Mastodon, X, Facebook, Instagram, or TikTok) DMs.