One of the major challenges in detecting deepfakes in the Global South is the quality of media produced in these regions. Unlike high-quality media that deepfake detection tools are usually trained on, cheap Chinese smartphone brands dominate the market in regions like Africa. These smartphones offer stripped-down features and produce much lower quality photos and videos. As a result, detection models often struggle to accurately identify manipulated media from these regions, leading to false positives and negatives.
Background noise in audio or video, as well as compression for social media, can further complicate the detection of deepfakes in the real world. Even minor alterations to media can result in faulty detection by sensitive models. The free tools available to journalists, fact checkers, and civil society members are often inaccurate when dealing with the lower quality material produced in the Global South. This inequity in representation in training data poses a significant challenge in detecting manipulated media accurately.
In addition to deepfakes, cheapfakes or media manipulated through misleading labels or simple editing techniques are prevalent in the Global South. These cheapfakes can be mistakenly flagged as AI-manipulated by faulty models or untrained researchers. There is a risk that tools more likely to flag content from outside the US and Europe as AI generated could lead to policy repercussions, creating imaginary problems that legislators might crackdown on. Inflating the numbers of AI-generated content could have serious implications at a policy level.
Building, testing, and running detection models require access to energy and data centers, which are not readily available in many parts of the world. Without local alternatives, researchers in the Global South are left with few options – paying for expensive off-the-shelf tools, using inaccurate free tools, or seeking access through academic institutions. The lack of computational resources makes it nearly impossible to run detection models locally. This reliance on external resources further delays the verification process, allowing potential damage to spread before content is confirmed.
While detection of deepfakes is crucial, focusing solely on this aspect may divert funding and support away from organizations and institutions that contribute to a more resilient information ecosystem. Instead of prioritizing detection, funding should be directed towards news outlets and civil society organizations that can foster public trust. Building a sense of credibility and trust in information sources is essential for combating misinformation effectively. However, the current allocation of funds may not be prioritizing this critical aspect of information integrity.
The challenges in detecting deepfakes in the Global South are multidimensional and complex. From the quality of media produced to the limitations in resources and tools for detection, there are significant barriers to identifying manipulated content effectively. Addressing these challenges requires a multifaceted approach that involves improving detection tools, enhancing access to resources, and supporting institutions that build trust in information dissemination. Only through concerted efforts can the Global South navigate the complexities of combating manipulated media and safeguarding the integrity of information.
Leave a Reply
You must be logged in to post a comment.