Kryptos

Summary

A copper sculpture in the courtyard of CIA headquarters carries about 870 encrypted characters in four passages. Three were broken in the 1990s. The fourth, ninety-seven characters long, has resisted the CIA, the NSA, and thirty-five years of the world's best professional and amateur cryptanalysts.

In September 2025 two researchers obtained its plaintext. They did not break the cipher. They found the answer in a box at the Smithsonian, because the artist had accidentally donated it.

Two months later the artist's archive, including the solution, sold at auction for $962,500 to a buyer, revealed in June 2026 as the investment firm Paradigm, which intends to keep it secret.

The cipher is still unbroken. That distinction is the whole of this record.

What is documented

The object. Kryptos, an S-shaped copper screen flanked by petrified wood, made by the artist Jim Sanborn and dedicated at Langley on 3 November 1990. The encryption was designed with Edward Scheidt, retired chairman of the CIA's Cryptographic Center. According to the CIA, 1,735 letters are cut into the copper. About half form the four encrypted passages; the rest form a Vigenère table. K1 and K2 share one method, the Vigenère, with different keywords, and K3 is a transposition.

K1, K2 and K3. All three were broken, and the sequence in which is itself part of the story.

An NSA team solved them by June 1993. The agency said so only after the public solution of 1999, without details; documents released in 2013, in response to a Freedom of Information Act request, confirmed the date.

A CIA analyst, David Stein, solved them in 1998, working with pencil and paper.

Jim Gillogly, an independent computer scientist, solved them and announced it publicly in 1999, with computer assistance. His was the first public solution, and the newspaper report announcing it also disclosed Stein's.

The plaintexts include a passage of Sanborn's own composition, a paraphrase of Howard Carter's account of opening Tutankhamun's tomb, and a line, in K2, reading that something is buried out there somewhere.

K4. Ninety-seven characters. Unbroken for thirty-five years.

The clues. Sanborn released fragments of the K4 plaintext over the years to keep the hunt alive: BERLIN in 2010, CLOCK in 2014, and further fragments in 2020. They did not crack it.

The cost to the artist. Sanborn has said publicly that the puzzle has destroyed marriages, brought strangers to his door, and provoked threats against his life. One person has written to him every week for twenty years. He began charging fifty dollars per submitted solution simply to reduce the volume.

The auction. In 2025, aged seventy-nine and in poor health, Sanborn announced that he would auction his complete Kryptos archive, including the K4 solution, and pass the role of keeper to somebody else. He said he no longer had the physical, mental or financial resources to hold it.

The discovery, which is not a solution. In early September 2025, two independent researchers, Jarett Kobek and Richard Byrne, obtained photographs of Sanborn's donated papers at the Smithsonian's Archives of American Art. Byrne took them, at Kobek's direction.

Among them were five pages of scrambled text that Sanborn had inadvertently included in his donation roughly two years earlier. They were copies of documents he had prepared in 1990 to demonstrate to the CIA's historical intelligence staff that the sculpture's text was not offensive.

From these, the researchers derived the K4 plaintext.

Sanborn's reaction, by his own account, was to put his head in his hands. The Smithsonian sealed his papers for fifty years. The researchers stated, in October 2025, that they do not intend to release what they found.

The sale. On 20 November 2025 the archive sold at RR Auction for $962,500, roughly double the high estimate. The buyer, anonymous at the time of the sale, announced on 12 June 2026 that it is the investment firm Paradigm. It has been designated the new keeper and has undertaken to preserve the secret. It says it arranged never to see the plaintext itself, and it has set up a system that checks submitted K4 solutions against a cryptographic hash of the answer.

K5. Sanborn has stated that Kryptos was designed to unravel in layers, and that a further passage exists which he calls K5. It has never been published, and the researchers who obtained the K4 plaintext have no knowledge of it.

Leading explanations

There is nothing anomalous here to explain. There is a distinction to protect, and it is the reason this record exists.

K4 has not been solved. Sanborn's own formulation is exact: it has been discovered, not solved. Elonka Dunin, one of the leading independent authorities on the sculpture, put it precisely: "One thing is to have the words. It's another to have the method."

Three people can now read a passage that nobody has decrypted. The encryption system for K4 has never been published. Sanborn has said that even Scheidt does not know the K4 solution; the method is recorded in the archive that Paradigm bought for $962,500.

This matters practically, not just semantically. A cipher is broken when somebody can demonstrate the transformation from ciphertext to plaintext. Nobody can. If a second message were encrypted with the K4 system tomorrow, not one person on Earth outside that small circle could read it.

The registry notes a source caution. The clearest statements of the discovered-not-solved distinction have come from the auction house selling the archive, which has an obvious commercial interest in the solution retaining value. The registry states the distinction anyway, because it is cryptographically correct regardless of who is making it, and because the independent researchers who obtained the plaintext have said the same thing themselves.

Current status

Partially explained. K1, K2 and K3 are solved. K4's plaintext exists, is known to a handful of people, and is not public. K4's cipher is unbroken and its method is a trade secret. K5 has never been published. In June 2026 Paradigm, the buyer of the archive, said it plans to release K5 and opened a system that checks proposed K4 solutions without revealing the answer.

The registry keeps this record for a distinction that this case makes unusually clean, and that most of the archive depends on: knowing the answer is not the same as knowing why it is the answer.

Somebody could have told you the plaintext of the Zodiac's Z340 at any point in the last fifty years, and it would have proved nothing. Oranchak, Blake and Van Eycke broke it, and that is a different thing entirely.

Last reviewed: September 2026. Records are provisional. Where the evidence changes, the entry changes. Found an error? Tell us.