At a time when artificial intelligence is increasingly used to proliferate sophisticated fake news and propaganda, Rareș Oancea, a 10th-grade student from Iași, Romania, has developed an innovative solution to combat the phenomenon. His platform, “TruthSeeker,” utilizes AI to analyze online articles, images, and videos, moving beyond binary true-or-false labels to provide comprehensive context. By scrutinizing sources, linguistic patterns, and manipulative techniques, the platform empowers users to better evaluate the credibility of the information they consume. Despite being a student of language and literature rather than computer science, Oancea was driven to create this tool after witnessing the intense wave of digital disinformation that skewed public perception during the 2024 Romanian presidential campaign.
The impetus for TruthSeeker arose when Oancea observed a surge of online content designed to create an almost divine image of extremist political figures, coupled with fabricated stories ranging from heroic rescues to absurd war-time tall tales. Initially, he suspected that these narratives were driven by automated “bots,” but his research revealed a more sobering reality: real people were spreading these falsehoods, often reacting with hostility toward opposing viewpoints. Realizing that the disinformation environment was deteriorating daily, he decided he could no longer remain a passive observer. Instead of attempting to argue with misled individuals, he spent over 2,000 euros of his own savings and hundreds of hours developing a system that encourages critical thinking rather than replacing it.
A standout feature of TruthSeeker is its forensic approach to visual media. Unlike many commercial large language models, the platform can analyze images to determine the likelihood of digital manipulation by assessing lighting, shadows, and perspective. Its efficacy was demonstrated when it successfully debunked viral claims that a photo showing a Romanian politician near Donald Trump at the Vatican had been digitally altered. Furthermore, the platform proved its value by analyzing emotive videos disseminated on Telegram; while the video descriptions claimed Ukrainian soldiers were harassing civilians, the AI analyzed the background audio, revealing that the soldiers were actually asking the civilians if they were safe. This capability allows the platform to trace the source of such content, identifying common patterns linked to specific disinformation-heavy channels.
Despite its successes, Oancea is candid about the limitations of his technology. He acknowledges that AI is not infallible and that one of his primary hurdles is avoiding “omission errors,” where the model might miss a relevant perspective from an established publication. Moreover, he faces an uphill battle against the sheer speed at which fake news travels, which studies suggest spreads six times faster than accurate reporting. To address concerns regarding AI trust, the platform does not simply issue a verdict; it provides links and citations, enabling users to verify the analysis themselves and ensuring the results are not merely “hallucinations” produced by the underlying technology.
The project faces significant operational challenges, primarily regarding the high costs of server maintenance and technical infrastructure required to train the AI. Having funded the project independently thus far, Oancea is now seeking external financing to sustain and scale his work. He remains firm that the platform will remain accessible to the public, rejecting the idea of placing his fact-checking tool behind a paywall. His vision for the future is to ensure that TruthSeeker serves as a robust resource for the general public, journalists, and institutions alike, acting as a collaborative tool that generates comprehensive graphs of connections between entities and misinformation networks.
Looking ahead, the next phase for TruthSeeker involves the development of a mobile application and a browser extension designed to provide real-time, on-screen verification of social media feeds and web articles. These tools aim to place an objective fact-checking assistant directly into the browsing experience of users, though the rollout is expected to take 12 to 18 months due to funding requirements and iOS system restrictions. By providing these tools, Oancea hopes to foster a more informed digital landscape, positioning his project not as a competitor to existing manual fact-checking organizations, but as an essential, high-tech complement in the ongoing global fight against the hybrid threat of disinformation.