return Home

Can machine learning detect “fake news”?

Francesco Marconi ()

While artificial intelligence can help journalists build a consistent fake news detector, it can also empower others to disseminate and even create new forms of misinformation, writes Francesco Marconi (@fpmarconi), manager of strategy and development at Associated Press.

Fake news is nothing new. The Roman Emperor Augustus led a campaign of misinformation against Mark Antony, a rival politician and general. The KGB used disinformation throughout the Cold War to enhance its political standing. Today fake news continues to serve as a political tool around the world, and new technologies are enabling individuals to propagate that fake news at unprecedented rates.

One of those new developments, artificial intelligence (AI), can help journalists build a consistent fake news detector, but AI can also empower others to disseminate and even create new forms of misinformation. To understand how, we need to take a quick detour and explain machine learning — one of the most important sub-domains of artificial intelligence

Detecting fake news with machine learning

Machine learning is, in the most basic sense, a system that learns from its actions and makes decisions accordingly, and it relies in turn on a process called deep learning which breaks down a single complex idea into a series of smaller, more approachable tasks. 

Machine Learning 1 ()

Thus, conceptually, machine learning can help detect fake news! An intelligent system that takes news stories as its input and a big ol' 'Fake' or 'Not fake' sticker as output. 

Machine Learning 2 ()

Machine learning (and deep-learning) relies mostly on algorithms, a set of rules that when followed leads to a desired output. But constructing algorithms is exceptionally difficult and the results can be catastrophic, especially when we rely on them to determine what news stories should be broadcast to our readers.

Algorithms make mistakes 

The two most common errors in this sort of machine learning are terms that we borrow from statisticians — Type I (false negative) and Type II (false positive) errors.

Machine Learning 3 ()

A false negative would mean that your machine labels a fake news item as not fake. We don’t want that. 

A false positive means your machine labels a real news story as fake. We don’t want that, either.

What we want is a system that can, with a high level of accuracy, label fake news as being fake, and real stories as not being fake. Again, we ask: How? 

We use news data to teach our algorithm.

Using news articles to train the AI

AI machines can best make decisions like what is and is not fake news when you define fake and real news — you do so by showing the machine tens of thousands of examples of each.

Machine Learning 4 ()

For that, you will need a data set of high quality journalism, as well as another collection with a sample of fake news, which could be sourced from a predetermined list of fake news.

The AI journalist

Remember, algorithms are written by humans, and humans make errors. Therefore, our AI machine may well make an error, especially in its early stages.

It’s the modern journalist’s job to know what her system is doing in order to be confident in what she is publishing.

There’s also an editorial decision to be made here. No system is going to be 100 per cent accurate, so which would you rather tend towards, false positives or false negatives? Would you rather a fake story be labelled as real or would you rather have a real story labelled fake?

Machine Learning 5 ()

Algorithmic errors aside, AI can help detect fake news. But as we mentioned earlier AI can also help disseminate it (all the more reason to understand AI, then!)

A new wave of “Fake-news”?

If you’ve worked in the journalism industry long enough you’ve probably been fooled by a doctored video, photo or sound bite. And every day the technology available to produce those fake news items is becoming easier to use and more publicly accessible. For instance, Adobe recently announced an AI project that is able to replicate the same tone of voice by simply analysing a sample of a speech, while a project developed by Stanford University researchers enables the manipulation of someone’s face in a video in real time.

Machine Learning 6 ()

In other words, the same sorts of machine learning and sub-domains of AI that can be used to fight fake news can also be used by others to propagate new types of misinformation.

The conclusion: Journalists need to understand AI.

Francesco first published this article on LinkedIn. It is reproduced with his permission here.

About Francesco:

Francesco Marconi is responsible for strategy and corporate development at the Associated Press, where he is part of the strategic planning team, identifying partnerships opportunities and guiding media strategy. Francesco complements his professional activity with academic research at Columbia University's Tow Center for Digital Journalism, where he is an Innovation Fellow.

Francesco studied business and journalism at the University of Missouri and completed his post graduate work as a Chazen Scholar at Columbia Business School’s Media Program. In 2014 he joined Harvard University’s Berkman Center for Internet and Society as an affiliate researcher studying the impact of data in journalism. Francesco started his career at the United Nations researching science and technology solutions for developing countries, resulting in the publishing of his first book and TED talk on Reverse Innovation.

More like this

How data and artificial intelligence are changing publishing

What will Artificial Intelligence mean for journalism?

Lack of trust in media: 'magazine media could offer a blue print out'

A behaviour expert’s 3 suggestions to cut through echo chambers and win trust

  • How can publishers thrive in challenging times?

    [Sponsored] Recently the WoodWing team traveled to London for the FIPP World Congress. For those of you who haven’t been lucky enough to attend yet, the FIPP World Congress is the largest and most high profile media event in the world. It brings together the world’s leading multi-platform media publishers and industry suppliers, to explore the latest trends and solutions.

    25th Oct 2017 Opinion
  • Are digital editions dead?

    Digital editions have been around for a long time, going all the way back to the late 90's. But in 2010 when the iPad hit the digital runway, publishers jumped on the tablet bandwagon faster than they could shout, “Hallelujah!”. The struggling publishing industry had found itself a saviour.  

    16th Oct 2017 Opinion
  • Publishers should re-emphasise their behavioural roots to take on Facebook and Google

    With Facebook and Google predicted to take half of the World’s total digital ad-spend in 2017, it’s no surprise that other players in the industry have raised concerns. But by updating their own data offerings to better reflect advertisers needs, media owners can keep pace with changing digital trends.

    25th Aug 2017 Opinion
  • How brands can capitalise on the experiential revolution

    If I were to ask you to describe the Internet of Things (IoT), I expect many of you would start to talk about how new technology is revolutionising the internet, providing “anything connectivity” through advanced networks, sensors, electronics, and software. And you wouldn’t be wrong.

    24th Aug 2017 Opinion
  • The state of brand licensing around the world

    Recently, there has been a period of time where there was somewhat of a slow-down in international brand activity as companies focused on shoring up their bases. However, this year we have seen an increasing number of reports surfacing about media companies adopting a more global outlook again – at least in certain segments. Does this mean a renewed focus on brand licensing, and in what form? And what is the outlook as we head into 2018?

    13th Nov 2017 Features
  • CDS Global and Zeddit announce tech partnership to help publishers grow print magazine subscriptions

    CDS Global and Zeddit announced a strategic technology partnership in the UK and Australia to provide advanced subscriber conversion capabilities for print magazine publishers. The partnership will focus on improving the conversion of visitors to magazine websites into subscribers for CDS Global clients.

    13th Nov 2017 Industry News
  • How Egmont is reaping rewards from creating

    One of the biggest drives for publishers in the past decade or two have been transitioning their print content to digital. For some it is all about maintaining the magazine's brand essence online, yet others have enjoyed success in amalgamating print publications to create new web first brands.

    13th Nov 2017 Features
  • Focus on behaviour, not technology, to identify emerging trends – Future Today Institute’s Amy Webb

    The Future Today Institute recently published its 2018 report into the emerging tech trends that are likely to shape the publishing industry in 2018. Here, we speak to Amy Webb, the founder of the organisation, about the development of study, and explains how better scientific modelling undertaken today can help us to predict future technologies.

    20th Nov 2017 Features
  • Advertising in the era of Google and Facebook

    The Facebook and Google duopoly are creating huge shifts in the allocation of advertising budgets. What does this mean for publishers, brands, and agencies? A panel of industry experts gave some answers during a recent panel discussion at the FIPP World Congress.

    20th Nov 2017 Features


Visit our Youtube channel



FIPP newsletters allow you to keep up with industry trends, research, training and events across the world



Get global coverage of your launches, company news and innovations


Upcoming @ FIPP

What’s happening now, what’s coming next

Go to Full Site