The paper explores the ethical issues surrounding deepfake technology, particularly the ways it distorts consent and privacy in the age of artificial intelligence, and how it can interfere with democracy, media credibility and trust among the masses generally. It also examines the ways in which deepfakes are used to spread political and other sorts of misinform, manipulate elections, and facilitate gender-based violence. Deepfakes are AI-generated or AI-manipulated audio, images, or videos created using deep learning techniques to produce highly realistic but fabricated media. .Initially, they were mainly used by individuals as a form of entertainment and research, but they have become a potentially devastating power within a short time.
Deepfakes technology is based on the advancements in Generative Adversarial Networks (GANs) proposed by Ian Goodfellow in 2014. GANs are models of AI intelligence that learn using massive datasets to generate images, videos, and voices which are very realistic. The increasing availability of free deepfake tools has heightened concerns because they can facilitate identity theft, fraud, misinformation, and other forms of cybercrime.
Deepfakes present substantial ethical challenges, such as the spread of misinformation, violations of consent, breaches of privacy, and reputational damage. These technologies facilitate the production of highly realistic but fabricated digital content, undermining society’s capacity to differentiate between authentic and false information. Increasingly, deepfakes are used to manipulate public opinion, perpetrate identity fraud, extort individuals, and generate non-consensual intimate imagery, with women being disproportionately targeted and significant human rights issues arising as a result.
Securing informed consent is critical when artificial intelligence is employed to generate or alter the images or likenesses of living individuals. While numerous privacy laws prohibit the unauthorized use of personal data, these regulations have not kept pace with the rapid development of generative AI technologies. The European Union’s General Data Protection Regulation (GDPR) may serve as a valuable legislative model because it recognises the “right to be forgotten,” enabling individuals, under certain circumstances, to request the removal of personal information from online platforms. This principle could inspire legal reforms aimed at mitigating the harmful effects of malicious deepfakes. Deepfakes may also be utilized to create false personal and business communications, enabling identity theft, financial scams, and various other types of fraud. These risks highlight the pressing necessity for enhanced legal protections and improved technical detection systems. Deepfakes represent a major risk to democratic governance by disseminating political misinformation, altering public perception, and undermining faith in democratic institutions. Fabricated videos or audio recordings during election campaigns can mislead voters and compromise electoral integrity.
The media sector is encountering greater challenges in fighting against misinformation created by AI. With the rise of realistic synthetic media, journalists and audiences face challenges in differentiating genuine content from invented material. As a result, trust in credible news outlets diminishes, rendering communities more susceptible to false information and propaganda.
Confidence is essential to democracy. The use of deepfake technology to create false evidence for journalism or legal processes greatly undermines trust in digital evidence. Creating and using dependable AI-driven forensic detection tools is vital for safeguarding the legal system and maintaining the integrity of public discussions.
Political disinformation is some of the most harmful applications of deepfakes. A tailored video of an Indian politician went viral during the 2019 elections, highlighting how deepfakes can affect public opinion and electoral conversations. Comparable worries have emerged in the United States, where AI-generated political material has escalated apprehensions regarding election meddling. Bangladesh also encounters similar dangers regarding its upcoming parliamentary elections. The danger goes far beyond Bangladesh. Elections across the globe have experienced an increase in AI-created false information and altered media. While there is scarce evidence that deepfakes have directly influenced election results, their growing sophistication poses a significant risk in the future. Bangladesh presently lacks specific laws aimed at combating malicious deepfakes, which complicates legal enforcement efforts.
The Digital Services Act (DSA) of the European Union signifies one of the initial extensive legislative initiatives to tackle online misinformation by placing increased obligations on digital platforms concerning illegal and AI-generated material. Comparable regulatory strategies might help Bangladesh tackle the increasing challenges brought by deepfake technology.
Some of the most concerning uses of deepfake technology are linked to gender-based violence. A significant issue is the production and spread of non-consensual deepfake pornography, where a person’s face is digitally placed onto explicit material without their awareness or permission. Such false content can inflict significant psychological harm, reputational injury, and social stigma on the victims. In South Korea, the rampant distribution of deepfake porn aimed at celebrities and everyday women led activists to call for tougher legal actions in 2019. Comparable issues have arisen in Bangladesh, where organizations advocating for women’s rights have documented cases of blackmail, defamation, and online harassment enabled by deepfake technology. While the Women and Children Repression Prevention Act of 2000 tackles cyber harassment, it fails to explicitly acknowledge or govern deepfake crimes resulting from advancements in artificial intelligence, thus creating a notable gap in the current legal structure.
The other jurisdiction that has adopted a serious approach to deepfakes is the UK, where the Online Safety Act 2023 literally makes it a criminal offense to create and share images or videos generated by deepfakes without consent. Bangladesh should consider adopting comparable legal reforms to provide effective protection for victims and ensure accountability for perpetrators.
Although there are various laws concerning cybercrime, defamation, and digital offenses, the present legal and regulatory system in Bangladesh is insufficient to tackle the specific challenges presented by deepfake technology. The swift progress of artificial intelligence has revealed numerous legislative and institutional shortcomings that obstruct the efficient prevention, detection, and prosecution of offenses related to deepfakes. The main deficiencies in the current framework consist of the following:
Lack of Clear Criminalisation: Bangladeshi law does not explicitly criminalise the malicious creation or dissemination of deepfakes, leaving significant gaps in prosecution.
Absence of AI-Specific Legislation: Bangladesh currently lacks a comprehensive legal framework governing the misuse of artificial intelligence technologies.
Law Enforcement Challenges: Law enforcement agencies often lack specialised AI forensic tools, technical expertise, and adequate training to investigate deepfake-related offences effectively.
Corporate Accountability: There is currently no comprehensive regulatory framework requiring AI developers and digital platforms to prevent or mitigate the misuse of deepfake technologies.
Limited Public Awareness: Public understanding of AI-generated content remains limited, increasing vulnerability to misinformation, fraud, and social engineering attacks.
Mashuk Abedin Chamak is a law graduate from Bangladesh University of Professionals.