December 15, 2006 (Computerworld) -- On Oct. 10, Florida's Orange County Comptroller's office completed an 18-month project designed to remove personally identifiable information from images of official records posted on its Web site.
The $750,000 effort began in April 2005 and involved the review of over 30 million pages in more than 12 million public records for items such as Social Security numbers (SSNs), bank account information and credit card numbers. In the end a total of 777,635 pages --2.6% of the total reviewed were found to have personal data and were redacted.
It's not entirely clear how many documents, some of which date back to 1970, still might contain personally identifiable information, said Carol Foglesong, assistant comptroller of Orange County. "There's going to be something we missed," she conceded. "But I think we got 99%," of the items that needed to be removed, she said.
Orange County's efforts are being replicated across dozens of counties in the state and around the country as local governments scramble to pull down documents from their Web sites or black out personal data from images of title deeds, tax liens, court papers and other public records.
As reported by Computerworld earlier this year, such images often contain personal identifiers and usually are accessible to anyone with Internet access. That has made county Web sites a veritable treasure trove of information for identity thieves, according to privacy advocates.
Many county governments still have not begun to address the prevalence of personal data despite the heightened public concerns about heightened said B.J. Ostergren a privacy advocate in Richmond, Va. In most cases, such sites continue to leak all sorts of sensitive personal data to anybody with Internet access. In some cases, county and state governments charge for access to the information, but even then the fees are relatively nominal compared to the value of the data, she said.
And many states are continuing to post SSNs in so-called Uniform Commercial Code (UCC) documents on their public Web sites, she said. UCC documents are filed with the state by banks and other creditors when an individual takes out certain types of loans.
But a growing number do appear to be attempting to fix the problem, she added.
"The more publicity this has gotten the more people are getting to know about the issue," she said. "I think a lot of people are beginning to put the skids on this sort of stuff" she said.
In October, for example, the council that oversees Washington's King County which includes Seattle passed an ordinance requiring the recorders office to remove online access to all title deed documents. The vote followed the discovery by a council member of more than 200 SSNs, including those of several public figures and athletes, in title deed documents on the county Web site.
Fears that the ordinance would hurt the county's business with mortgage companies and others prompted some resistance from the county recorder's office initially, said a spokesman for Regan Dunn, the councilman who sponsored the bill. But the title deeds have been pulled down from the site and won't be restored until the Social Security Numbers are somehow blocked from public view, the spokesman said.
In another example, the recorder's office in Grant County, Indiana, pulled all of its document images from the Internet in July after a lawsuit related to identity theft was filed against the county. "There are no definite plans to put them back up on the Internet although Social Security Numbers will be redacted starting next year," said County Recorder Dixi Fischer Conner.
Laws in several states, including Florida, New York and Washington, require recorders to redact personal data from online records. But removing such information can be a huge challenge because of the sheer number of documents that need to be examined.
Foglesong said that in Orange County each image was reviewed once by redaction software from Mentis Technology Solutions LLC and then manually checked by workers. "We learned that software combined with a human review is much more dependable than human eyes searching page after page," she said.
Software-only approaches are also unreliable said Dana DeBeauvoir, clerk of Travis County in Texas. "It's definitely not the automated process that software vendors will have you believe," she said, adding that workers have to double-check documents to verify they need to be redacted and to ensure proper redactions have taken place. The wide range of document formats and the fact that some documents are handwritten can pose problems for redaction software applications, DeBeauvoir said.
"The more publicity this has gotten the more people are getting to know about the issue," she said. "I think a lot of people are beginning to put the skids on this sort of stuff" she said.
In October, for example, the council that oversees Washington's King County which includes Seattle passed an ordinance requiring the recorders office to remove online access to all title deed documents. The vote followed the discovery by a council member of more than 200 SSNs, including those of several public figures and athletes, in title deed documents on the county Web site.
Fears that the ordinance would hurt the county's business with mortgage companies and others prompted some resistance from the county recorder's office initially, said a spokesman for Regan Dunn, the councilman who sponsored the bill. But the title deeds have been pulled down from the site and won't be restored until the Social Security Numbers are somehow blocked from public view, the spokesman said.
In another example, the recorder's office in Grant County, Indiana, pulled all of its document images from the Internet in July after a lawsuit related to identity theft was filed against the county. "There are no definite plans to put them back up on the Internet although Social Security Numbers will be redacted starting next year," said County Recorder Dixi Fischer Conner.
Laws in several states, including Florida, New York and Washington, require recorders to redact personal data from online records. But removing such information can be a huge challenge because of the sheer number of documents that need to be examined.
Foglesong said that in Orange County each image was reviewed once by redaction software from Mentis Technology Solutions LLC and then manually checked by workers. "We learned that software combined with a human review is much more dependable than human eyes searching page after page," she said.
Software-only approaches are also unreliable said Dana DeBeauvoir, clerk of Travis County in Texas. "It's definitely not the automated process that software vendors will have you believe," she said, adding that workers have to double-check documents to verify they need to be redacted and to ensure proper redactions have taken place. The wide range of document formats and the fact that some documents are handwritten can pose problems for redaction software applications, DeBeauvoir said.
Since public records laws prohibit recorders from altering originals in any form, it was also necessary to make copies of the images before making redactions, she said. This can impose an additional storage burden which, while small, still needs to be factored into costs, she said.
Travis County removed all document images from its Web site in June this year because of identity theft concerns and started putting them back online last week after redacting sensitive information. DeBeauvoir said about 11 million documents had been redacted and restored another five million newer records will go back online soon, she said. "I would give the quality of the work an A+," she said.
Finding the resources needed for the task can be a big problem, said Sue Baldwin director of the Broward County records division in Florida. "The main challenge has been dedicating sufficient personnel to review the 'probably qualifying' documents queued up by the [redaction] software," she said.
To meet that challenge, the Secretary of State's office in Missouri hired three part-time workers and four staffers. The team has been working since summer to redact Social Security numbers from about 75,000 UCC documents on its Web site. A spokeswoman for the Secretary of State's office said that so far about 20,000 records have been redacted; The plan going forward is to bring in an "electronic redaction system" to help complete the task, she said.
But the success of redaction efforts remains in question said David Bloys a retired private investigator who publishes a newsletter called "News for Public Officials" in Shallowater, Texas.
Bloys said he checked Travis County's Web site last week "and the very first document I found on their Web site contained social security numbers, drivers license information and a home address on the first page," he said. Subsequent pages in the document provide details about the individual's financial, family and medical information, he added. What's worse is that the Web site had numerous pages like that, Bloys claimed. there are numerous such pages still up on the county Web site. "Redaction just doesn't work," Bloys said. "I think the only way to really protect these documents is to make sure they stay within the four walls of the court house," he said.
