The pervasive spread of disinformation has been exacerbated by the Internet, with social media platforms becoming a breeding ground for misleading information. A Eurobarometer report revealed that 28% of EU citizens have fallen victim to fake news, a crisis compounded in nations with restricted press freedom. The rapid evolution of generative AI tools, such as ChatGPT for text and MidJourney for images, has raised concerns about the potential intensification of disinformation. Laurence Dierickx from the University of Bergen warns that the accessibility and speed of AI-generated content dissemination have reached unprecedented levels, contributing to misinformation’s rampant proliferation.
The dark side of AI amplifying disinformation
While AI has the potential to produce amusing content like AI-generated images of the Pope in a puffer jacket or deepfakes depicting Donald Trump’s arrest, malicious actors are increasingly harnessing AI’s capabilities for their sinister agendas. Anja Bechmann from Aarhus University points out that AI has enabled the creation and amplification of misinformation worldwide, often employing techniques like deepfakes from platforms like DALL.E and MidJourney. This phenomenon of “rapid disinformation attacks” involves unleashing false information swiftly and broadly to create immediate chaos.
The challenge of tackling AI-generated nonsense
Fact-checkers face an uphill battle when confronted with the overwhelming task of manually verifying AI-generated content. Overworked and often subject to online harassment, these professionals struggle to keep up with the speed and volume of disinformation spread. The need for an efficient solution has become more pressing than ever.
Ironically, the same technology that fuels disinformation can also be harnessed to combat it. AI can play a crucial role in detecting and mitigating the issue. An effective approach involves deploying end-to-end content detection models that compare statements against a predefined database to ascertain their veracity. Meta, the parent company of Facebook, has introduced Sphere, an AI tool that cross-references claims with the collective knowledge base of Wikipedia. However, limitations arise when relying on Wikipedia’s accuracy as a reference, and questions linger about the reliability of such an approach in journalism.
Fact-checking collaborations
Combatting disinformation necessitates collaboration between tech giants and the news industry. Project Origin, a collaboration involving Microsoft and media organizations like the BBC and The Telegraph, seeks to preserve the integrity of news dissemination from creation to consumption. This initiative employs various tools to identify accurate content using metadata from transcoded files.
Similarly, NODES, a project involving AFP, CNRS, and others, focuses on analyzing the journey of disinformation across traditional and digital media in multiple European languages. This initiative exemplifies the synergy between fact-checkers and news organizations, a partnership crucial for tackling disinformation effectively.
A Norwegian model: AI-powered fact-checking
Norway offers a compelling example of cooperation between fact-checkers and news outlets. Faktisk.no, a non-profit fact-checking organization, collaborates with six competing news organizations in Norway. Their AI department, Faktisk Innsikt, employs databases, Natural Language Processing, and Norwegian Language Models to combat disinformation in the country.
Fostering comprehensive solutions
The fight against disinformation extends beyond individual efforts. FERMI, a consortium of 17 institutions funded by the European Union, is actively working on multiple initiatives to counter disinformation and fake news. However, the transition to full automation for fact-checking with AI remains a distant goal. The current workflow necessitates the combined efforts of AI and human fact-checkers.
As the news industry contemplates integrating AI into newsrooms, it’s essential to leverage AI tools not only for content creation but also to counteract the spread of disinformation. A multi-pronged approach involving AI, collaborative fact-checking efforts, and strengthened partnerships between tech companies and news organizations is crucial to create a more informed and trustworthy digital landscape.
