What’s Up With Twitter’s Content Moderation Policies?

Rosemary Shahan is a consumer advocate who’s been at the forefront of auto policy for nearly four decades. Through her organization Consumers for Auto Reliability and Safety (CARS), she has run successful campaigns to make vehicles safer and prevent used-car company rip-offs. A board member at the Consumer Federation of America, Shahan was responsible for California’s first auto lemon law (which became a national model), the federal airbag mandate, and much more.

Right now, Shahan is locked out of her Twitter account, over a claim of “abusive behavior.” That behavior involves two tweets where Shahan is engaging in the same public advocacy she has been doing her entire career. The tweets specifically target AutoNation, a Fortune 500 used-car chain with over 300 locations nationwide. AutoNation never asked Twitter to take the tweets down, a spokesperson confirmed to the Prospect. So what motivated Twitter’s action?

When Donald Trump had his tweets slapped with warning labels and then had his account suspended over false claims about the presidential election, liberals cheered. Finally, Twitter was instituting its terms of service and shutting down contrary behavior. But speech policing is inherently subjective, and in the case of Shahan and others, it can be used as a weapon to block advocates from using essential communications platforms to speak out against political and economic power.

More from David Dayen

Tools to suppress voices never work in only one direction. Historically, they have been used to stop dissent. It’s why we have to be incredibly careful with letting tech platforms that have no real expertise or even interest in content moderation engage in these practices without transparency, and why we have a bigger problem with the private companies that have become essential means of communication in the first place.

Shahan’s two tweets, now removed, were posted to the CARS Twitter feed (@cars4consumers) in early January. They amplify an issue that she has been fighting for years. In late 2016, after Donald Trump’s election victory, AutoNation resumed a policy of selling used vehicles with unrepaired, or “open,” safety recalls. New cars cannot be sold with open recalls, but there is no parallel law on the books for used vehicles.

AutoNation had voluntarily prohibited this practice, with CEO Mike Jackson saying in February 2016, “We feel the time has passed that it’s appropriate to take a vehicle in trade with a significant safety recall and turn around the next day and sell it to consumers.” But Trump’s election led AutoNation to abruptly reverse course. A story in Automotive News quoted Jackson bluntly stating that, thanks to Trump’s election, “there’s no way that that issue is going to be addressed from a regulatory point of view.” In other words, the expectation of a deregulatory administration meant there was no longer any need to sell cars that would be ready to drive safely.

This saved AutoNation the expense of finding parts, some of which are scarce, and doing repairs before sales. But it puts drivers at risk, as Shahan wrote in the tweets, which have been removed from Twitter:

Your donation keeps this site free and open for all to read. Give what you can…


The reference to “open season on used car buyers” is certainly tough rhetoric. And the tweets included a picture of Stephanie Erdman, who was severely injured when a Takata airbag exploded upon impact. Erdman has testified before Congress about her injury, and the photo was entered into evidence at that hearing. “While it’s graphic, sometimes that’s what it takes to help people understand what’s at stake,” Shahan told me.

This, then, is a topic that already had generated public attention and news coverage. For her part, Shahan has posted about this issue repeatedly, as you’d expect from a consumer advocate. A Twitter search yields 14 different tweets on the subject, including one from back in 2018 that used the “open season” rhetoric. She is currently working with Sen. Richard Blumenthal (D-CT) and auto safety groups on legislation that would close the loophole on open recalls. It’s not a conspiracy theory; it’s an area of public policy.

But a few weeks after the postings, Twitter took action. The site’s support team informed Shahan that the tweets were removed and her posting privileges suspended, because the tweets violated their “abusive behavior” policy, which Twitter defines as “an attempt to harass, intimidate, or silence someone else’s voice.” The only entity referred to in the postings is AutoNation, so Twitter is effectively saying that one consumer advocate with a feed is trying to silence the voice of a company with $3.5 billion in profits last year.

“I’ve shown the tweets to other auto safety advocates and asked them if they thought I had crossed over any lines,” Shahan said, “and they all agreed that the tweets, while strongly worded, seem appropriate under the circumstances.” Nevertheless, Twitter locked her out. Shahan appealed, and Twitter responded that “Our support team has determined that a violation did take place, and therefore we will not overturn our decision.”

Shahan cannot access Twitter unless she manually removes the tweets, and Twitter would display a message on them saying they were removed “because they violated Twitter’s policies against ‘abusive behavior,’ without anyone being able to see what they were and judge for themselves,” Shahan explained. For a consumer advocate with powerful opponents, having to state on a public forum that she was abusive could be wielded in a way to sap Shahan’s credibility.

But how did Twitter’s support team know about these tweets in the first place?

I asked AutoNation if they flagged the tweets for Twitter. Spokesperson Marc Cannon said he was unaware of the whole situation, and AutoNation had nothing to do with it. “I know Rosemary very well, and quite frankly if I had a problem with Rosemary I’d pick up the phone and call her,” Cannon said. “I don’t need anybody else to help me.” While Cannon said he appreciates Twitter taking down openly hostile and brutal comments from AutoNation customers, he agreed that this policy advocacy didn’t rise to that level.

The decisions Twitter makes about moderation are fairly opaque, from how they sift through millions of tweets and decide which ones to flag, to what due diligence goes into the appeals process, to what distinctions they make between “abusive” content and aggressive advocacy. Twitter did not respond to my questions on these and other issues.

Your donation keeps this site free and open for all to read. Give what you can…


The situation resembles one that hits substantially closer to home. This week, labor reporter Mike Elk tweeted about a story from the Prospect which featured an interview with Josh Brewer, who is the lead organizer of the unionization effort at an Amazon warehouse in Bessemer, Alabama. Brewer said in the interview that Amazon has been holding “classroom” sessions with workers, and union supporters are singled out in front of the class. Elk supplied a screenshot of that paragraph.

According to Twitter user @TimothyS, the tweet was initially hidden for containing “suspicious content.” (It can be viewed now.) Steven Greenhouse, the highly regarded labor reporter who conducted the interview for us, has reported on other workers telling him the same story about anti-union meetings. What in the content was suspicious? Twitter again has not responded.

Elk told me that other reporters working on the Amazon union drive story, including for his own Payday Report, have also had tweets flagged as “suspicious.” More Perfect Union, an advocacy media institution, told the Prospect that it had a video of Amazon organizers in Alabama pulled by Twitter after racking up 500,000 page views in a few hours.

In the absence of a response from Twitter, we’re left wondering about their content moderation policies and how they are applied. Even if for every left-leaning news outlet or consumer advocate being flagged there’s some other tweet from the right also being flagged, the way in which this process unfurls may well appear to the user as arbitrary and confusing. But you can certainly see a way that companies could flag posts that put them in an unfavorable light and get Twitter to suppress those comments.

There’s a bigger problem here. Twitter has every right to manage content on its platform. But the platform, and others like Facebook, have become global town squares that are key to the dissemination of information. Moderation in that context is a difficult task, especially for tech companies that have always been so thoroughly uninterested in enforcing terms of service.

We should not be outsourcing speech policies to private companies. And we never should have embraced laissez-faire competition policies that created a situation where a small handful of these companies have become essential communications platforms that are necessary conduits of news and information. Maybe network effects render that impossible; maybe we had no choice other than this narrowing in social media. But we certainly haven’t tried any alternatives to prevent these power dynamics from taking hold.

Instead, we have advocates being silenced for bizarre reasons, news sites—including this news site!—being flagged as “suspicious,” and a completely obscure standard for speech on one of the more critical sites for attention and awareness. It’s a completely inappropriate system with few options for solutions, unless and until we begin to break the dominance of Big Tech.