May 09, 2001, 2:42 PM — Scientists attending the Fourth Information Hiding Workshop last month in Pittsburgh viewed demonstrations of how to conceal information from repressive regimes and how to build watermarks to track documents, and they got a lesson in how to use vague threats of a lawsuit to muzzle academic discussion.
The normally uncontroversial conference focused on exploring ways to slip extra bits of information into unexpected locations. The science of hiding information, often called steganography, is gaining attention because copyright holders, especially those in the music industry, hope to use the hidden bits of information to tag and even corral digital versions of songs, books, movies and other works of art.
The tools for creating these "digital watermarks" were developed by members of the Secure Digital Music Initiative and released in a highly controlled contest that publicly challenged others to test their strength.
But when a group of scholars from Princeton University, Rice University and Xerox Corp.'s Palo Alto Research Center (PARC) discovered ways to circumvent those tools for embedding copyright information in songs and planned to present a detailed paper at the workshop, tensions surfaced. The Recording Industry Association of America sent a letter encouraging the researchers to keep the information secret and noting that they could face legal action.
The group withdrew the paper and issued a statement saying, "Litigation is costly, time-consuming and uncertain, regardless of the merits of the other side's case."
The conference was devoted to describing new and enhanced ways to hide or lock information. Another group from Xerox PARC demonstrated tools for hiding information in the background of text documents. One approach used small hash lines oriented in different directions to encode the information; others used the size and width of the characters. Both technologies let people embed digital signatures into text documents.
Many researchers wrestled with the problem of how to create watermarks that could be controlled by encryption keys. Such a system would work like public-key encryption (PKI) algorithms: Only one person could embed the watermark, but anyone could test its presence. This type of tool would allow companies to embed watermarks that act as digital signatures in electronic documents. Another approach used neural networks as keys, and a third method was based on zero-knowledge proofs, which allow a user to verify that he has the information without revealing it.
Other techniques used private keys, so the information in the watermark could be recovered only if the key was known. These techniques are more useful for hiding information in a way that can't be identified. Some papers described sophisticated techniques that eluded detection by ensuring that the hidden information was statistically identical to the surrounding data, providing camouflage.