Congress Holds Oversight Hearing On Data Mining Systems

The Subcommittee heard testi-mony from Steve McCraw, Assistant Director, Office of Intelligence, Federal Bureau of Investigation (FBI); Admiral James L. Loy, Director, Transportation Security Administration (TSA); and Dr. Anthony Tether, Director, Defense Advanced Research Projects Agency (DARPA).

Mr. McCraw began his testimony by asserting that the FBI¹s own data systems contain information that is legally and lawfully collected, and that new data mining systems, like the agency's SCOPE project, will allow analysts to search the agency's existing databases for links, associations, and relationships among individuals. McCraw conceded that the FBI's systems rely in part on data compiled by public sector companies that are not always accurate. He testified that follow-up investigations are often necessary to confirm the information. Rep. William Clay (D-MO) challenged McCraw on the FBI's recent decision to lift the data accuracy requirements for the agency's largest criminal justice database, the NCIC (see EPIC Alert 10.07). In response, McCraw emphasized the strict guidelines governing the use of NCIC. He also agreed to review the matter further.

Admiral Loy also sought to reassure the Subcommittee about the accuracy of the TSA's proposed CAPPS-II system that will identify passengers for additional screening before boarding a plane. He testified that the TSA would establish a Passenger Advocate to investigate passengers' concerns about being identified for pre- screening, but he conceded that the investigation could take some time and that it would not always be possible to inform passengers of the reasons for the additional screening. Admiral Loy asserted that unlike the classic definition of "data-mining," CAPPS-II would result in a "traveler-activated" search. The traveler's provided name, address, telephone number, and date of birth would be used first to authenticate the traveler's identity through public sector databases, and then would be run through government data systems and assessed a risk threat score. Admiral Loy did not specify the risk assessment techniques, but announced that a new Federal Register notice would be promulgated soon based on comments the agency received from its first notice on the program.

Similarly, Dr. Tether used the hearing as an opportunity to address public concerns about DARPA's Total Information Awareness program (TIA). He contrasted TIA with traditional data mining techniques that comb through large amounts of information to detect previously unnoticed correlations. He testified that DARPA is not pursuing such techniques and instead is developing a different approach to research. He said that the approach begins with the development of a hypo- thetical attack scenario, and then leads to the use of data mining to discover whether patterns of information correlated with that scenario actually exist. He stressed that this process would decrease the threat of erroneously flagging innocent activities and persons as suspicious, and emphasized that audit techniques would ensure that data is not used for unauthorized purposes. When representatives inquired whether TIA would use consumers' transaction information held by private companies, Dr. Tether sought to distance the agency from such potential uses, stating that a researcher hired by DARPA may be contemplating such practices, but the agency had not yet made any such formal plans.

Documents obtained by EPIC through an Freedom of Information Act request, however, reveal that one of TIA¹s goals is to develop "innovative technologies to architect, populate and exploit" reposi- tories "for combating terrorism." Repositories were defined as "a new kind of extremely large, omni-media, virtually-centralized, and semantically rich information repository that is not constrained by the limited commercial database products available today."

The Subcommittee plans to reconvene in two weeks to examine the privacy and civil liberties questions raised by the programs.

Source: EPIC's letter submitted for a hearing on Data Mining, Current Applications and Future Possibilities.