The OED's first citation is 1569, Thomas Norton attacking Catholics for flying a false flagge of religion, purely figurative. The literal naval sense, enemy ships under friendly colors, appears in the 1800s; pirates entered the story later. Naval law actually permitted false colors with one rule: hoist your true flag before firing. Documented modern cases include Mukden 1931, Gleiwitz 1939, the Nazi-staged Polish attack that opened the invasion, and Operation Northwoods, the 1962 Pentagon proposals, never executed, to stage Cuban-attributed attacks as pretext for war.
Every documented case, Gleiwitz, Mukden, Northwoods, made the next suspicion more reasonable, and the internet industrialized that suspicion. Now the term is deployed within hours of any atrocity, usually meaning the opposite of its origin: not a hidden hand revealed by evidence, but evidence refused in favor of a hidden hand. Britannica notes the concept has been co-opted to explain away tragedies that challenge worldviews.
The claim: because Northwoods was proposed, signed by the Joint Chiefs, and only stopped at the President's desk, the category unthinkable officially lost meaning, and every subsequent denial inherits the doubt. Verdict: Northwoods is declassified and real, and it does shift the burden of proof permanently. But one proposed-and-rejected plan is evidence that review worked as much as that plotting happened. The sane position is the uncomfortable one: false flags are real, rare, and provable, which is why proof is the price of the accusation.
If documented false flags exist, does that license suspicion of every flag?