The Karush-Kuhn-Tucker (KKT) conditions are necessary conditions for a solution in nonlinear programming to be optimal, given that certain regularity conditions are satisfied. They generalize the method of Lagrange multipliers to handle inequality constraints, providing a set of equations and inequalities that must be satisfied at the optimal point.