Roughly 2 million tweets peddled conspiracy theories about the coronavirus over the three-week period when the outbreak began to spread outside China, according to an unreleased report from an arm of the State Department, raising fresh fears about Silicon Valley’s preparedness to combat a surge of dangerous disinformation online.

The wrongful, harmful posts floated a number of hoaxes — suggesting, for example, that the coronavirus had been created by the Bill and Melinda Gates Foundation or was the result of a bioweapon. These and other identified falsehoods represented 7 percent of the total tweets the government studied and were “potentially impactful on the broader social media conversation,” according to the report, which was obtained Saturday by The Washington Post.

The Global Engagement Center, the propaganda-fighting program at the State Department whose name appears on the document, said it focused its analysis on countries excluding the United States between Jan. 20 and Feb. 10, a period during which the World Health Organization declared the novel coronavirus an international health emergency. In total, the Global Engagement Center explored 29 million foreign posts, the report said.

Some of the misinformation exhibited “evidence of inauthentic and coordinated activity,” according to the report, raising the specter that foreign governments or other malicious actors may have deliberately tried to sow fear and discord about the international health emergency — much as Russian agents had done during the 2016 presidential election in the United States.

But the report did not detail fully what led to that conclusion, nor did it attribute the information to a specific government source. Previously, agency officials signaled in public news reports that some of the coronavirus-related content on social media may be tied to agents of the Kremlin. But the State Department has provided no evidence of that to tech giants or the public, and it does not mention Russia once in the document obtained by The Post.

Nor does the State Department report include a list of suspicious accounts or much other description as to why it thought certain tweets were part of an inauthentic campaign in the first place.

The State Department did not respond to a request for comment. Representatives at Twitter declined to comment on the report, but said the company was working with federal officials on coronavirus.

The data still lay bare the broad challenges facing Twitter and its tech peers as the novel coronavirus spreads globally. Already under siege by a raft of half-truths and falsehoods, Silicon Valley has taken a more aggressive approach to the deadly outbreak, seeking to vet or remove a wider array of harmful content than it has in the past.

“This analysis is a window into how false narratives about coronavirus are truly global and spread faster than the virus itself,” said Graham Brookie, the director of the Atlantic Council’s Digital Forensic Research Lab, which tracks disinformation. “It highlights that the medical and information challenge presented by coronavirus don’t respect our neatly defined borders.”

In recent weeks, Facebook, Google and Twitter introduced policies and features designed to direct users searching for information about the outbreak to authoritative sources, including the Centers for Disease Control and Prevention. Facebook, in particular, also has sought to ban posts, photos and videos that push harmful misinformation, including cures that do not exist.

Such steps have won tech giants plaudits from critics who long have clamored for the industry to take a more active role in policing a wide array of harmful content on the Internet. But it has not curtailed the flood of coronavirus myths and conspiracy theories entirely: Private Facebook groups, and hard-to-find but still popular YouTube videos, continue to push inaccuracies about the malady and its origins.

It may still take months for researchers and other experts to fully assess Silicon Valley’s effectiveness, but the data included in the State Department’s unclassified dispatch, dated Feb. 19, offers at least an initial glimpse into how a panicked globe so far has turned to social media.

The top tweets during the 20-day period that government researchers studied reflected a mix of people seeking more information about Wuhan — the disease epicenter in central China — talking about censorship in the tightly controlled country, joking about the disease and making racist connections between Asians and the spread of the coronavirus, the Global Engagement Center found.

One of the most widely shared articles in that period wrongly correlated the outbreak with “people eating bat soup,” a racist trope, though the State Department said some of the most shared links included official resources, such as an online hub from Johns Hopkins University tracking the contagion.

Some of the most dangerous false narratives, however, appear to have been bolstered on Twitter by inauthentic means. The hoax attributing coronavirus to the Gates Foundation, for instance, benefited from “amplification by inorganic activity and inauthentic accounts,” the report said, including one user who since has been suspended by Twitter. Some of the tweets linked to YouTube videos, according to the State Department document, suggesting that misinformation troubles may be broad in scope.

Still another hoax said an arm of the Defense Department created the virus and targeted China. Those tweets gained traction with the aid of accounts that appeared to be automated, known as bots, that have tweeted at an unnatural rate in recent weeks, according to the State Department's conclusions.

Critically, though, the government’s study found that traditional media sources served as an important counterweight to a small number of false, yet still popular, websites that peddled fake narratives. One of the most widely read articles in Brazil, with more than 13 million views, sought to dispel fears about the coronavirus, the State Department said.