Introduction
Social Identity Theory (SIT), developed by Tajfel and Turner (1979), offers a valuable lens for understanding group dynamics within organisations, emphasising how individuals derive part of their self-concept from group memberships. This theory involves processes such as social categorisation, where people classify themselves and others into groups; social identification, adopting group norms; and social comparison, seeking positive distinctiveness for one’s group (Turner and Oakes, 1986). In modern organisations, particularly in the technology sector, these psychological bonds can override formal structures, as seen in the 2023 leadership crisis at OpenAI. This essay applies SIT to analyse the events surrounding the abrupt dismissal and reinstatement of CEO Sam Altman, highlighting tensions between non-profit safety-focused identities and for-profit accelerationist ones. By examining conformity, decision-making, conflict, diversity management, remote teams, and power dynamics, the essay argues that strong social identities can undermine hierarchical governance in knowledge-based firms. The analysis draws on SIT to demonstrate how identity conflicts led to governance failure, while proposing strategies for resolution. As a student studying Social Identity Theory, this case illustrates the theory’s practical relevance in contemporary organisational behaviour.
SIT, Conformity Studies, and Obedience in the OpenAI Crisis
Traditional theories of organisational authority, such as Weber’s (1947) concept of legitimate power, suggest that directives from a board should be followed without question. However, the OpenAI crisis, where the board dismissed Altman on 17 November 2023, only to face a mass employee revolt and reverse the decision days later, challenges this view. SIT provides a more insightful explanation, focusing on identification-based conformity (Hogg and Turner, 1987). Employees did not rebel against authority per se but conformed to the norms of their in-group, which positioned the board as an out-group.
The employee response was swift: over 700 of 770 staff signed an open letter demanding Altman’s return, threatening to join Microsoft if unmet. This can be seen as a reaffirmation of in-group identity, where Altman embodied the group’s prototype – innovative, ambitious, and successful (Hogg, 2001). The board’s actions threatened this identity, intensifying conformity pressures. Indeed, not signing the letter risked social exclusion, aligning with Reicher, Haslam, and Smith’s (2012) reinterpretation of obedience experiments, which views resistance as collective action to protect shared identity rather than mere defiance.
From a student’s perspective in SIT, this event underscores how conformity is not blind obedience but tied to group salience. The crisis revealed that in high-stakes tech environments, psychological bonds can rapidly mobilise collective action, overriding formal power.
SIT and Group Decision Making
The board’s opaque decision-making process exemplifies SIT’s insights into groupthink and insulation (Janis, 1982). Comprising members aligned with Effective Altruism and AI safety, the board formed a cohesive in-group perceiving threats from rapid AI development. This led to out-group stereotyping, viewing Altman’s commercial faction as reckless, and created an echo chamber by excluding key stakeholders like Microsoft.
SIT attributes this to depersonalisation, where board members prioritised their ‘safety guardian’ prototype over organisational realities (Turner, 1991). They assumed broad support for their decision, underestimating employee loyalty to Altman. This false consensus effect highlights a limitation in SIT: while it explains in-group biases, it requires integration with other theories for full predictive power in crisis scenarios. Critically, the board’s ideological consistency proved organisationally detrimental, illustrating how isolated identities can lead to poor governance.
As someone studying SIT, I note that this case shows the theory’s strength in analysing decision biases but also its need for contextual adaptation in fast-evolving sectors like AI.
SIT, Conflict, Group Polarisation, and Conflict Resolution
Intergroup conflict at OpenAI stemmed from competing identities: ‘Doomers’ emphasising existential risks and ‘Boomers’ (accelerationists) focusing on rapid innovation. Following ChatGPT’s 2022 success, the accelerationist group gained status, threatening the safety-focused non-profit identity (Tajfel and Turner, 1979). This created an ‘us vs. them’ dynamic, exacerbated by structural silos that promoted group polarisation (Moscovici and Zavalloni, 1969). Safety teams viewed product releases as dangerous, while product teams saw safety as bureaucratic.
Resolution partially followed the Common Ingroup Identity Model (CIIM) (Gaertner et al., 1993), with Altman’s reinstatement and a new board integrating safety, capital, and innovation interests. However, for long-term stability, OpenAI should foster a superordinate ‘Responsible Innovation’ identity, recategorising members into a unified group that values both speed and safety. This could reduce bias by emphasising shared goals, such as building AGI responsibly.
Studying SIT, I appreciate how CIIM offers practical tools for conflict resolution, though its success depends on leadership buy-in, which was initially lacking here.
SIT and Managing Diverse Groups and Teams
OpenAI’s structure – a non-profit board overseeing a capped-profit entity – created faultlines, dividing the organisation along ideological lines (Lau and Murnighan, 1998). This values-based diversity proved more disruptive than demographic diversity, leading to identity incompatibility and conflict (Ashforth and Mael, 1989). The non-profit group felt marginalised as the for-profit arm dominated, prompting the board’s assertive action.
Effective management requires cross-categorisation, highlighting overlapping identities like ‘scientists’ or ‘innovators’ to blur divides (Crisp and Hewstone, 2007). The crisis exposed leadership failures in addressing these faultlines, allowing psychological divisions to align with structural ones. Arguably, proactive identity management could have prevented escalation.
From an SIT learning viewpoint, this highlights the theory’s applicability to diversity, though it sometimes overlooks how power imbalances influence identity salience.
SIT and Managing Multi-Cultural Remote Groups and Teams
Despite OpenAI’s San Francisco base, the revolt unfolded digitally via Slack and X (formerly Twitter), involving a distributed workforce. SIT’s Social Identity Model of Deindividuation Effects (SIDE) explains this: in computer-mediated settings, reduced personal cues amplify group norms, fostering conformity (Reicher, Spears, and Postmes, 1995). Employees saw unanimous signals like emojis and posts, creating perceived consensus against the board.
This digital dynamic amplified in-group salience, with the board’s silence allowing employee narratives to dominate. For remote teams, leaders must engage digitally to shape identities. Typically, such environments can enhance inclusivity in multi-cultural groups, but here they accelerated polarisation.
As a student, I find SIDE particularly relevant to modern work, though it may undervalue cultural differences in digital conformity.
SIT and Understanding Power in Organisations
The crisis revealed referent power’s dominance over legitimate power in knowledge economies. The board held legal authority but lacked prototypicality, while Altman embodied the group’s optimistic identity (Hogg, 2001). Employees’ threats to leave demonstrated that power resides in human capital, not legal structures.
SIT suggests leadership depends on embodying group values; the board’s out-group status subverted their authority. This necessitates rethinking governance, where boards must align with workforce identities.
Studying SIT, this case illustrates power as identity-contingent, with implications for AI governance.
Conclusion
The OpenAI crisis exemplifies how SIT explains governance failures through identity conflicts, from conformity and polarisation to power shifts. It shows that in tech firms, social identities can dismantle hierarchies, urging superordinate identities for stability. Implications include the need for identity-aware leadership in AI, balancing innovation and safety. This analysis, as a student of SIT, reinforces the theory’s value while highlighting its limitations in fully capturing rapid organisational change.
Self-Reflection on Team Experience
Using Gibbs’ Reflective Cycle (Gibbs, 1988), I reflect on my team experience applying SIT to the OpenAI case.
Description: Our team collaborated on this essay, researching the crisis timeline and SIT applications via a shared Kanban board and video calls. I focused on digital aspects and SIDE.
Feelings: Excitement from the topic’s relevance turned to anxiety over blending theory with current events, and frustration arose from debates on focus.
Evaluation: Information gathering succeeded, but integrating conflict models caused delays. Peer review strengthened theoretical accuracy.
Analysis: Task conflict (Jehn, 1995) improved outcomes, teaching me to embrace disagreement for better analysis.
Action Plan: Future projects will start with concept mapping to resolve differences early.
(Word count: 1624, including references)
References
- Ashforth, B. E., & Mael, F. (1989). Social identity theory and the organization. Academy of Management Review, 14(1), 20-39.
- Crisp, R. J., & Hewstone, M. (2007). Multiple social categorization. Advances in Experimental Social Psychology, 39, 163-254.
- Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A., & Rust, M. C. (1993). The common ingroup identity model: Recategorization and the reduction of intergroup bias. European Review of Social Psychology, 4(1), 1-26.
- Gibbs, G. (1988). Learning by Doing: A guide to teaching and learning methods. Oxford: Further Education Unit.
- Hogg, M. A. (2001). A social identity theory of leadership. Personality and Social Psychology Review, 5(3), 184-200.
- Hogg, M. A., & Turner, J. C. (1987). Intergroup behaviour, self-stereotyping and the salience of social categories. British Journal of Social Psychology, 26(4), 325-340.
- Janis, I. L. (1982). Groupthink: Psychological studies of policy decisions and fiascoes (2nd ed.). Boston: Houghton Mifflin.
- Jehn, K. A. (1995). A multimethod examination of the benefits and detriments of intragroup conflict. Administrative Science Quarterly, 40(2), 256-282.
- Lau, D. C., & Murnighan, J. K. (1998). Demographic diversity and faultlines: The compositional dynamics of organizational groups. Academy of Management Review, 23(2), 325-340.
- Moscovici, S., & Zavalloni, M. (1969). The group as a polarizer of attitudes. Journal of Personality and Social Psychology, 12(2), 125-135.
- Reicher, S. D., Haslam, S. A., & Smith, J. R. (2012). Working toward the experimenter: Reconceptualizing obedience within the Milgram paradigm as identification-based followership. Perspectives on Psychological Science, 7(4), 315-324.
- Reicher, S. D., Spears, R., & Postmes, T. (1995). A social identity model of deindividuation phenomena. European Review of Social Psychology, 6(1), 161-198.
- Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33-47). Monterey, CA: Brooks/Cole.
- Turner, J. C. (1991). Social influence. Buckingham: Open University Press.
- Turner, J. C., & Oakes, P. J. (1986). The significance of the social identity concept for social psychology with reference to individualism, interactionism and social influence. British Journal of Social Psychology, 25(3), 237-252.
- Weber, M. (1947). The theory of social and economic organization. New York: Oxford University Press.
