Supporters of Islamic State could control as many as 90,000 Twitter accounts worldwide, a new study suggests.
The terror group is able to "exert an outsized impact on how the world perceives it" because of its use of social media and number of online followers, according to the Washington-based Brookings Institute report.
It recommends that governments and social media companies work together to find new ways to tackle the problem of accounts supporting the Islamists, who have posted gruesome pictures and videos purporting to be of executions in Iraq and Syria online, as well as propaganda rhetoric.
It argues that the problem also applies to other social networks and extremist groups such as far-right racist organisations.
The report, by Brookings academic JM Berger and technologist Jonathon Morgan, says: "While we do not believe that any mainstream social media platform wishes to see its services used to further acts of horrific violence, we also suspect some would rather not be bothered with the challenge of crafting a broad and coherent response to the issue.
"While we can sympathise with the challenges and dilemmas such a response would entail, it is clear that social media companies do feel an obligation to respond to some social standards and illegal uses of their services.
"We are not aware of any major company that takes a hands-off approach to the use of its platform to promote child pornography or human trafficking - or, less dramatically, phishing, spam, fraud, and copyright violations.
"Extremism, while raising thornier issues, merits attention, especially when faced with a rising challenge of violent groups who manipulate platforms to reap the rewards of spreading images of their cruelty."
The study of IS-linked accounts between September and December estimated there were between 46,000 and 70,000 IS-supporting Twitter accounts, with the researchers believing that the true figure was towards the lower end of this scale but setting an absolute maximum at 90,000.
Their analysis was based on "robust" data collected about 50,000 accounts, and partial information about a further 1.9 million.
Only a small number of the accounts were able to have their locations identified, as most had this function switched off. But of those that could be located, the vast majority were in the Middle East and North Africa. Other were found in the UK, France, Spain, Italy, Switzerland, Belgium and Australia, but these numbers were in single figures, the report found.
They noted that platforms including Facebook and YouTube have already introduced changes aimed at tackling extremist material.
Twitter had started suspending accounts linked to IS - also known as Isis - by the time the research was started, but the authors said this created a new risk, arguing: "While suspensions appear to have created obstacles to supporters joining Isis's social network, they also isolate Isis supporters online.
"This could increase the speed and intensity of radicalisation for those who do manage to enter the network, and hinder organic social pressures that could lead to deradicalisation.
"Further study is required to evaluate the unintended consequences of suspension campaigns and their attendant trade-offs. Fundamentally, tampering with social networks is a form of social engineering, and acknowledging this fact raises many new, difficult questions.
"Social media companies and the US government must work together to devise appropriate responses to extremism on social media.
"Although discussions of this issue often frame government intervention as an infringement on free speech, in reality, social media companies currently regulate speech on their platforms without oversight or disclosures of how suspensions are applied."