A New Assistant in the Newsroom
In newsrooms and verification hubs around the world, a new, uniquely 21st-century assistant is being put to work: generative AI. Rather than replacing journalists, these powerful tools are being used to augment their abilities. Fact-checkers are experimenting
with large language models (LLMs) like ChatGPT and other specialised AI tools for a range of support tasks. These include summarising lengthy government reports, translating content from different languages, and analysing large datasets to spot emerging narratives or anomalies. For instance, an AI can quickly parse hours of political speeches to identify specific claims that warrant a closer look, saving valuable time. This allows human experts to focus less on tedious data collection and more on the critical work of verification.
The Need for Speed and Scale
The primary driver for this adoption is the overwhelming speed and scale of modern misinformation. False narratives can spread across the globe in minutes, far outpacing traditional, manual fact-checking methods. AI offers a way to level the playing field. By automating parts of the discovery process, fact-checking organizations can respond more swiftly. Some teams, like Georgia's MythDetector, use AI to find new instances of a claim that has already been debunked, helping to track and counter its spread. Similarly, the Norwegian organisation Faktisk has used AI to create informative maps that help verify images and videos from conflict zones. The goal is not automation for its own sake, but to make human-led verification more efficient and scalable in a chaotic information ecosystem.
Beyond Summaries to Deeper Research
The use of AI extends beyond simple summaries. These tools are also becoming brainstorming partners and research assistants. Fact-checkers can use AI to formulate questions, find related lines of inquiry, and uncover the history of a specific falsehood. Some platforms are specifically designed for academic and scientific research, helping journalists find and understand credible studies to support their checks. By asking an AI to analyze a piece of text for biased language or logical fallacies, a journalist can get a quick first impression of its potential credibility before diving deeper. This support allows for more thorough and contextualized fact-checks, moving beyond a simple true or false verdict to explain the bigger picture to the public.
The Human Remains the Final Judge
Despite the potential, there is a firm consensus among leading journalistic and fact-checking bodies: AI is a tool, not an oracle. Organizations like the International Fact-Checking Network (IFCN) and the Poynter Institute stress that any information generated by an AI must be treated as unvetted source material. Human oversight is non-negotiable. The risk of AI models 'hallucinating'—inventing facts with complete confidence—is significant. As such, no credible organization is using AI to make the final determination of whether a claim is true or false. The technology assists, but the final judgment, ethical reasoning, and accountability remain squarely in the hands of trained human professionals who can understand context, intent, and nuance in ways that an algorithm cannot.
Navigating a Tightrope of Risks
Adopting AI is a delicate balancing act. Fact-checkers are keenly aware of the paradox: they are using a technology that is a primary engine for creating sophisticated disinformation. There are significant ethical concerns, including the inherent biases present in AI training data, which can perpetuate stereotypes or skewed perspectives if not carefully managed. Furthermore, over-reliance on these tools could potentially dull the critical thinking skills of journalists over time, a concern highlighted in recent MIT research. In response, newsrooms are developing strict internal policies and guidelines for AI use, emphasizing transparency and ensuring that human editors and fact-checkers are involved at every stage of the process to maintain accuracy and public trust.
