|

BP & Sombrero Meme Deepfakes: Revisiting the Ethics of Synthetic Media in the Age of AI

by Cat Hubka & Athena C. Gordon | Xchanges 21.1, Fall 2026

Other Ethical Approaches

While virtue ethics has many benefits, scholars argue that it should not be the only framework for addressing ethical challenges because there are many benefits of combining multiple perspectives (Vallor, 2016; Colton & Holmes, 2018). Multiple ethical frameworks provoke us to reflect upon the nuances of different ethical situations and help us contextualize different ways that we can respond to complex ethical dilemmas. We believe that deontology and consequentialism, in addition to virtue ethics, can help us engage with synthetic media responsibly.

Deontology

Deontology is a normative approach to ethics that focuses on the ethical duty of an action, rather than its consequences. In other words, one’s actions must align with one’s ethical duty, and this alignment constitutes good decisions. This framework can be particularly useful when examining our responsibility as the TPC community to use AI ethically. Deontology has been used for hundreds of years as a guide for decision-making, often applied to religious duty and rationality during the European Enlightenment (Pihlaja, 2025). Immanuel Kant’s categorical imperative holds that acting in accordance with universal ethical duties, regardless of consequences or personal desires, is necessary for moral action. In her analysis of BP’s actions in response to the oil spill, McDonnell uses a deontological lens, focusing on duty, truthfulness, and respect for audiences, to argue that their manipulation of images was unethical because manipulating images with the purpose to deceive the public violates BP’s moral responsibilities. In addition, McDonnell suggests that if photo manipulation were considered a universal law, there would be extensive social harm from the deceptive framing of truth. Therefore, BP’s manipulation of images cannot be ethical because it cannot be justified as a universal law.

The development of AI guidelines, both regulatory and non-regulatory, can help center ethical decision-making with a deontological approach. On a functional level, guidelines prevent unethical action by defining what is acceptable and unacceptable in the use of AI. Creating AI guidelines also helps delineate common values within an organization and promotes accountability for members. Ethical guidelines are not a new practice in the TPC community, as many guidelines have been created to foster professional standards since the 1950s for organizations such as the Association of Technical Writers and Editors (TWE), the Society of Technical Communicators (STC), and the Association of Teachers of Technical Writing (ATTW) (Malone, 2011). Before writing or implementing AI ethical guidelines, the TPC community should consider the practical challenges of guidelines, such as ease of implementation, scope of audience, delineation between non-binding ethics and legally enforceable laws, and applicability to real user situations (Panarese et al., 2025; Ryan & Stahl, 2021; Smuha, 2019). For guidelines to be useful with consistently evolving technology, TPC communicators need to create an actionable code and enable flexibility to engage with new knowledge (Holmes et al., 2022; Smuha, 2019).

It’s important to note that developing AI guidelines is only a starting point for engaging responsibly with AI. Holmes et al. (2022) argue that developing a framework is “necessary but not sufficient” and requires further conversations and actions to promote ethics (p. 520). AI frameworks need to provide proactive foundational guidance for ethical actions, not just to prevent the community from unethical action, for this to be an effective intervention to address ethical issues of AI. In addition, the combination of other ethical approaches, such as creating organizational cultures grounded in virtue ethics, may be most effective to promote responsible decision-making with emerging AI technology.

Consequentialism

In contrast to deontology, consequentialism is a normative ethical approach that focuses on the consequences of an action. If an action results in a good outcome, then the action is justifiable and ethically acceptable. McDonnell is skeptical that the potential consequence of easing public pressure and promoting greater peace of mind was positive enough to justify BP’s photo manipulation. Further complicating this scenario, McDonnell (2015) writes, “there is no way to truly and accurately deem such an objective good or right,” suggesting that there are not many reasonable methods to evaluate BP’s doctoring of images through a consequentialist lens (p.8). There would need to be an understanding of what is “good” to be able to apply consequentialism to a scenario. McDonnell shows that even from a consequentialist standpoint, justifying deception is difficult because deciding what constitutes a “good” outcome is inherently subjective. The subjectivity in defining the quality of actions affects AI-related decision-making. The creators of the sombrero memes likely believed that they were doing a “good” because the development and distribution of the memes would influence lawmakers to end the government shutdown. However, popular public reaction demonstrated that the memes caused more harm than good. Clearly, there were competing conceptions of the ethical “good.”

Let’s imagine that we could define what “good” or “bad” consequences are. With the rapid evolution of technology, there are challenges in defining the potential consequences of AI. We already know some of the impacts of AI on our society. For example, AI can exacerbate racial, gender, and other biases throughout its design and use (Bender et al., 2021; Panarese et al., 2025). Large language models use large datasets and create “Stochastic Parrots,” in which bias is reproduced and amplified without human accountability. We understand that this bias from synthetic media can disproportionately affect historically marginalized communities (Colón Vargas, 2025), and it would be irresponsible for TPC practitioners and educators to use AI tools without recognizing their potential harms. On the other hand, there are many positive opportunities for AI, such as reducing repetitive work, making it easier to find information, and supporting mental health through affective computing (Carradini, 2024; Yadav & Maini, 2025). Given the multitude of potential outcomes, the TPC community would need to hold space for multiple consequences and try to weigh the net benefits of an action while making decisions.

While we have an understanding of some of the possible benefits and harms of AI, it’s impossible to fully understand the scope of the consequences of synthetic media to guide our decision-making through a consequentialist approach alone. One of the major concerns about AI technologies is the lack of understanding of how AI systems work, often likened to a “black box” (Katz, 2025; Siau & Wang, 2020). This “black box” is problematic because the AI system can evolve without human monitoring and produce information in ways humans do not understand, creating fear and eroding trust (Katz, 2025). This issue of the “black box” makes it difficult to use consequentialist frameworks in decision-making, as we do not fully understand the consequences. Considering these uncertain outcomes of AI, we believe that virtue ethics can provide a helpful framework to navigate rapid technological change and promote constructive decision-making. We can reflect upon the different possible consequences of AI, define the virtues that we aspire to embody, and develop responsible agency with synthetic media.