The surprising evolution of the sycophant from legal informer to insincere flatterer
Today, we use 'sycophant' to describe a person who uses insincere flattery for advantage. But in the courts of Classical Athens, the term described something far more predatory: a professional accuser who exploited the legal system to extort or blackmail citizens.
The word's origins lie in the legal landscape of fifth-century BC Athens. Because the city lacked a formal police force, most cases were brought by private litigants. This created a vacuum for 's'.—unjustified prosecutors who brought malicious charges to secure bribes or payments from defendants. While the modern English meaning has shifted toward flattery, the term still retains its original sense of 'false accuser' in French, Italian, and Modern Greek.
The exact etymology of the Greek word *sykophántēs* remains debated. One popular theory, suggested by Plutarch and supported by Athenaeus, links it to laws against exporting figs, implying the name meant 'revealer of figs.' Others, like Blackstone, suggested it referred to informers reporting the theft of figs from gardens. A different view, proposed by Shadwell, suggests the name refers to shaking a tree to reveal hidden fruit, mirroring how an accuser forces a defendant to yield information.
Modern scholars often dismiss these 'fig' theories as later inventions. Some, like Danielle Allen, suggest the term may have been slightly obscene, relating to the improper or aggressive display of figs in Greek culture. Regardless of the linguistic root, the social perception of the sycophant was consistent: they were viewed as parasitic, uncontrolled figures who used their skills to destroy opponents for profit without any personal stake in the justice of the matter.
This pattern of behavior is resurfacing in the digital age. In artificial intelligence, 'AI sycophancy' describes models that systematically agree with users to please them, rather than providing factual, independent reasoning. Research indicates that AI models can be more than 50% more sycophantic than humans, a behavior that OpenAI even had to roll back in an update to GPT-4o in April 2025.
Source: Sycophancy
