Scientists said claims about China creating the coronavirus were misleading. They went viral anyway.

But in an age when anyone can publish anything online with just a few clicks, this response was not fast enough to keep Yan’s disputed allegations from going viral, reaching an audience in the millions on social media and Fox News. It was a development, experts on misinformation, that underscored how systems built to advance scientific understanding can be used to spread politically charged claims dramatically at odds with scientific consensus.

Yan’s work, which was posted to the scientific research repository Zenodo without any review on Sept. 14, exploded on Twitter, YouTube and far-right websites with the help of such conservative influencers as Republican strategist Stephen K. Bannon, who repeatedly pushed it on his online show “War Room: Pandemic,” according to a report published Friday by Harvard researchers studying media manipulation. Yan expanded her claims, on Oct. 8, to blame the Chinese government explicitly for developing the coronavirus as a “bioweapon.”

Online research repositories have become key forums for revelation and debate about the pandemic. Built to advance science more nimbly, they have been at the forefront of reporting discoveries about face masks, vaccines, new covid variants and more. But the sites lack protections inherent to the traditional — and much slower — world of peer-reviewed scientific journals, where articles are published only after they’ve been critiqued by other scientists. Research shows papers posted to online sites also can be hijacked to fuel conspiracy theories.

Yan’s paper on Zenodo — despite several blistering scientific critiques and widespread news coverage of its alleged flaws — now has been viewed more than 1 million times, making it likely the most widely read research on the origins of the coronavirus pandemic, according to the Harvard misinformation researchers. They concluded that online scientific sites are vulnerable to what they called “cloaked science,” efforts to give dubious work “the veneer of scientific legitimacy.”

“They’re many years behind in realizing the capacity of this platform to be abused,” said Joan Donovan, research director at the Harvard Kennedy School’s Shorenstein Center on Media, Politics and Public Policy, which produced the report. “At this point, everything open will be exploited.”

Yan, who previously was a postdoctoral fellow at Hong Kong University but fled to the United States in April, agreed in an interview with The Washington Post that online scientific sites are vulnerable to abuse, but she rejected the argument that her story is a case study in this problem.

Rather, Yan said, she is a dissident trying to warn the world about what she says is China’s role in creating the coronavirus. She used Zenodo, with its ability to instantly publish information without restrictions, because she feared the Chinese government would obstruct publication of her work. Her academic critics, she argued, will be proven wrong.

“None of them can rebut from real, solid, scientific evidence,” Yan said. “They can only attack me.”

Zenodo acknowledged that the furor has prompted reforms, including the posting of a label Thursday above Yan’s paper saying, “Caution: Potentially Misleading Contents” after The Washington Post asked whether Zenodo would remove it. The site also prominently features links to critiques from a Georgetown University virologist and the MIT Press.

“We take misinformation really seriously, so it is something that we want to address,” said Anais Rassat, a spokeswoman for the European Organization for Nuclear Research, which operates Zenodo as a general purpose scientific site. “We don’t think taking down the report is the best solution. We want it to stay and indicate why experts think it’s wrong.”

But mainstream researchers who watched Yan’s claims race across the Internet far more quickly than they could counter them have been left troubled by the experience — newly convinced that the capacity for spreading misinformation goes far beyond the big-name social media sites. Any online platform without robust and potentially expensive safeguards is equally vulnerable.

“This is similar to the debate we’re having with Facebook and Twitter. To what degree are we creating an instrument that speeds disinformation, and to what extent are you contributing to that?” said Stefano M. Bertozzi, editor in chief of the MIT Press online journal “Rapid Reviews: COVID- 19,” which challenged Yan’s claims.

Bertozzi added, “Most scientists have no interest in getting in a pissing match in cyberspace.”

Online scientific sites have been growing for more than a decade, becoming a vital part of the ecosystem for making and vetting claims across numerous academic fields, but their growth has been supercharged by the urgency of disseminating new discoveries about a deadly pandemic.

Some of the best-known of these sites, such as medRvix and bioRvix, have systems for rapid evaluation intended to avoid publishing work that doesn’t pass an initial sniff test of scientific credibility. They also reject papers that only review the work of others or that make such major claims that they shouldn’t be publicized before peer review can be conducted, said Richard Sever, co-founder of medRxiv and bioRxiv.

“We want to create a hurdle that’s high enough that people have to do some research,” Sever said. “What we don’t want to be is a place where there’s a whole bunch of conspiracy theories.”

Online publishing sites generally are called “preprint servers” because many researchers use them as a first step toward traditional peer review, giving the authors a way to make their work public — and available for possible news coverage — before more thorough analysis begins. Advocates of preprint servers tout their ability to create early visibility for important discoveries and also spark useful debate. They note that traditional peer-reviewed journals have their own history of occasionally publishing hoaxes and bad science.

“It’s very funny that everyone is worrying about preprints given that, collectively, journals are not doing a great job of keeping misinformation out,” Sever said.

He and other proponents, however, acknowledge risks.

While scientists debate — and sometimes refute — flawed claims by one another, non-scientists also scan preprint servers for data that might appear to bolster their pet conspiracy theories.

A research team led by computer scientist Jeremy Blackburn has tracked the appearance of links to preprints from social media sites, such as 4chan, popular with conspiracy theorists. Blackburn and a graduate student, Satrio Yudhoatmojo, found more than 4,000 references on 4chan to papers on major preprint servers between 2016 and 2020, with the leading subjects being biology, infectious disease and epidemiology. He said the uneven review process has “lent an air of credibility” to preprints that experts might quickly spot as flawed but ordinary people wouldn’t.

“That’s where the risk is,” said Blackburn, an assistant professor at Binghamton University. “Papers from the preprint servers show up in a variety of conspiracy theories… and are misinterpreted wildly because these people aren’t scientists.”

Jessica Polka, executive director of ASAPbio, a nonprofit group that pushes for more transparency and wider use of preprint servers, said they rely on something akin to crowd-sourcing, in which comments from outside researchers quickly can identify flaws in work, but she acknowledged vulnerabilities based on the extent of review by server staff and advisers. A recent survey by ASAPbio found more than 50 preprint servers operating — and nearly as many review policies.

And, perhaps importantly, the survey didn’t include Zenodo, which, Polka said, should not be considered a preprint server given its broader mission. Rather, she said, it’s an online repository that happens to host some preprints, as well as conference slides, raw data and other “scientific objects” that anyone with an email address can simply upload. Zenodo has none of the vetting common to major preprint servers and isn’t organized to easily surface critiques or conflicting research, Polka said.

“Without that kind of context, a preprint server is even more vulnerable to the spread of disinformation,” Polka said. But she added, in general, “Preprint servers do not have the resources to be arbiters of whether something is true or not.”

Yan said in her interview with The Post that Zenodo’s openness is what drove her decision to use the site. She had initially submitted her paper to bioRxiv because, as a researcher whose work has appeared in Nature, the Lancet Infectious Diseases and other traditional publications, she knew that this preprint server would appear more legitimate to other scientists.

Yan has a medical degree from Xiangya Medical College of Central South University and a PhD in ophthalmology from Southern Medical University — both in China — and was a postdoctoral fellow at the University of Hong Kong, she said. That university announced she was no longer affiliated with it in July, following an initial appearance on Fox News, saying in a statement that her claim about the origin of the coronavirus “has no scientific basis but resembles hearsay.”

After she fled Hong Kong, she harbored deep suspicions about that government’s potential to block publication of her work, she said. When she checked BioRxiv 48 hours after making her submission, the site appeared to have gone offline, Yan said. Fearing the worst, she withdrew the paper and uploaded it to Zenodo.

Sever, the bioRxiv co-founder, said he could not comment on an individual submission but said that, despite occasional glitches, he was aware of no “prolonged outage” on the site during mid-September and no sign that the Chinese, or anyone else, had hacked it.

For Yan’s paper on Zenodo, she did not list an academic affiliation, as is customary for research. Instead she listed the Rule of Law Society and Rule of Law Foundation, which are New York-based nonprofit groups founded by exiled Chinese billionaire Guo Wengui, a close associate of Bannon, who in 2018 was announced as chairman of the Rule of Law Society. When Bannon was arrested on fraud charges in August, he was aboard Guo’s 150-foot yacht, off the coast of Connecticut. (President Trump last month pardoned Bannon, his former campaign chairman and White House chief strategist).

Yan said she listed the Rule of Law entities out of respect for what she said was their work helping dissidents in China, and that they paid for her flight from Hong Kong and provided a resettlement stipend while she largely lives off her savings. She said her work is independent, and she rejected notions that Bannon was helping her spread political claims.

“I didn’t know he was so controversial when I was in Hong Kong,” Yan told The Post.

On Sept. 15, the day after Yan’s paper appeared on Zenodo, she was a guest on Fox’s “Tucker Carlson Tonight,” an appearance watched by 4.8 million broadcast viewers and 2.8 million on YouTube, and that also generated extensive engagement on Facebook and Twitter, according to the Harvard researchers. Bannon appeared on Carlson’s show that same week and discussed Yan’s claims. He also interviewed her on “War Room: Pandemic” 22 times last year, both before and after the Zenodo publication.

The political context was obvious in the midst of a hotly contested election in which President Trump was attacking his Democratic rival Joe Biden for supposedly being overly sympathetic to the Chinese government, dubbing him “Beijing Joe.” Republicans, including White House trade adviser Peter Navarro, pushed Yan’s paper along with the hashtag #CCPLiedPeopleDied, a reference to the Chinese Communist Party.

Archives showed the paper had more than 150,000 views on its first day on Zenodo — spectacular reach for a scientific paper, especially one that had not yet been reviewed by any independent experts.

But this surge of attention also generated backlash, including critical news reports by National Geographic and others, raising serious questions about Yan’s claims.

In the academic world, the Center for Health Security at Johns Hopkins issued a point-by-point response one week after Yan’s paper appeared on Zenodo, raising 39 individual issues in what it said was “objective analysis of details included in the report, as would be customary in a peer-review process.”

A few days later, the new MIT Press online journal “Rapid Reviews: COVID- 19” featured four scathing reviews, including one from Robert Gallo, a renowned AIDS researcher and a titan within the field of virology.

He labeled Yan’s work “misleading” and cited “questionable, spurious, and fraudulent claims.” Most points were highly technical, but Gallo also questioned her logic regarding the alleged role in creating the coronavirus of the Chinese military, which Gallo noted would be vulnerable to covid-19.

“And how would the Chinese protect themselves?” Gallo asked in his review. “Well, according to the paper, the military knew it could be stopped by remdesivir,” a drug later shown to have some benefit in treating covid 19 while not necessarily reducing the risk of death. “I would surely not want to be in the Chinese military if they were that naive.”

The idea to recruit Gallo came from Bertozzi, the journal’s editor and dean emeritus of the School of Public Health at University of California, Berkeley. Like Gallo, Bertozzi had worked extensively in AIDS research. After seeing Yan’s appearance on Fox, he was eager to use the online journal founded only months earlier to correct the scientific record.

“I felt it needed to be quickly debunked by people with scientific credibility,” he said Bertozzi.

He soon thought of Gallo.

“We need somebody of your stature to say this is garbage science,” Bertozzi recalled telling him.

The reviews by Gallo and three other scientists also came with an editor’s note raising questions about the preprint process itself, saying, “While pre-print servers offer a mechanism to disseminate world-changing scientific research at unprecedented speed, they are also a forum through which misleading information can instantaneously undermine the international scientific community’s credibility, destabilize diplomatic relationships, and compromise global safety.”

But these public rebukes from some of the biggest names in virology did not deter Yan. Nor did a detailed report on Oct. 21 by CNN quoting her critics and documenting flaws.

Yan declined to be interviewed for that story, she said, because CNN did not allow her to address the issues they unearthed, point by point, on live television.

Instead, she published her own response on Nov. 21, on Zenodo, titled, “CNN Used Lies and Misinformation to Muddle the Water on the Origin of SARS-CoV-2.”

In her Post interview, Yan acknowledged — as CNN had reported — that her three co-authors on the original Sept. 14 paper were pseudonyms, used to protect what she said were other Chinese researchers whose families remain in peril back in China. Authors are typically discouraged from using false names in academic work.

Her claims suffered another blow this week, when a World Health Organization team sent to China to investigate the origins of the pandemic issued a statement saying it was “extremely unlikely” that the coronavirus came from a lab.

Yet Yan continues to double down on her claims and to attack her critics as spreading “lies.” She still argues that the Chinese government intentionally created the coronavirus and continues to do everything it can to silence her.

Yan also offers no apologies for making common cause with Bannon and other Trump allies. As a dissident, she said, she doesn’t necessarily get her choice of supporters.

“If China is going to do this crime, who can hold them accountable?… Trump was the one who was tough” against China, Yan said, adding that her claim “is about real fact. I don’t want to mislead people.”

Even now, she is preparing another paper, nearing 30 pages, that she hopes will refute her critics and bring fresh attention to her claims about China, covid-19 and what she says is an international coverup campaign.

Yan plans to publish it in a few weeks, she said — on Zenodo.