Whatever It Takes to Win: Rivalry Increases Unethical Behavior

GAICIL Insights

Competition and rivalry sound like synonyms.¹ Whatever It Takes to Win: Rivalry Increases Unethical Behavior shows they behave like different chemicals. Gavin J. Kilduff, Adam D. Galinsky, Edoardo Gallo and J. James Reade propose that people are more likely to act unethically when competing against rivals than against non-rival competitors. Across an archival study and a series of experiments, rivalry was associated with more unsporting behavior, more deception and greater willingness to use unethical negotiation tactics.

The paper digs beneath the association into the psychology. The data reveal a serial mediation pathway. Rivalry raises the psychological stakes of competition by increasing actors' contingency of self-worth and status concerns. Those stakes produce a stronger performance-approach orientation, which in turn increases unethical behavior. The effects of competition, the authors conclude, depend on the relationships and prior interactions between actors. The opponent's identity changes the game.

That is a precise refinement of how markets work. Ordinary competition disciplines firms and benefits consumers. The paper isolates a relational form, rivalry between actors with a history of competing against each other, that warps competitive psychology. Actors who feel rivalry, the authors write, are more likely to engage in the behavior that defines the ugly side of competition, such as deception, cheating and sabotage.

Now look at how the artificial intelligence race is usually described: a handful of leading firms, and a handful of nations, competing repeatedly, knowing each other intimately and attaching enormous status to leadership. That description resembles the paper's definition of rivalry, though the paper does not study AI firms or nations. The research suggests rivalries heighten psychological stakes and raise the temptation to cut corners. In AI, corner-cutting can look like overstated capabilities, rushed releases or massaged test results. The danger lies in the grudge match.

The remedy implied is elegant: keep the competition, defuse the incentive to cheat. Transparent benchmarks, independent evaluation and clear rules against deception shrink the payoff to unethical tactics. When the scoreboard is credible, gaming it pays less, and competitive energy flows into genuine product improvements. A trustworthy referee lets rivals play hard without playing dirty.

Organizations face the same chemistry internally. AI teams compete for resources and recognition, and firms can breed rivalries through ranking systems and winner-take-all rewards. Managers should notice how such structures shape ethical behavior. Incentives that reward performance without inflating the status stakes of beating a particular rival may preserve motivation while limiting misconduct. Leaderboards pitting two named teams against each other might be the most expensive motivational tool a firm ever uses.

Language matters at the national level too. Framing AI development as a zero-sum race against a named rival may raise exactly the psychological stakes the paper identifies. Framing it as broad competition to produce useful, reliable technology may preserve competitive pressure while dampening the rivalry that encourages deception. Rhetoric shapes incentives, and political speeches are, in this light, part of the incentive design of an industry.

The authors call rivalry a widespread, powerful and largely unstudied phenomenon with significant organizational implications. That invites research on rivalry in innovation: whether rivalries in research and development accelerate discovery, whether they increase misconduct, and how institutions can harvest the first while limiting the second. The answer could shape how prizes, grants and procurement competitions are structured.

The deeper lesson is that competition has psychological texture. The same pressure that drives effort can, under rivalry, encourage unethical shortcuts. Markets and institutions that make honesty observable and deception costly harness rivalry's energy while blunting its corrosive edge. As AI firms and nations compete intensely, building that infrastructure, credible measurement and transparency, is one of the most effective ways to ensure the race delivers innovation that can be trusted.

The most important safety investment in a heated technology race may be an independent, trusted scoreboard. Rivals who cannot fake the score have to win it.