The Government Accountability Office (GAO) revealed that the FBI did not appropriately disclose the database’s impact on public privacy until the FBI was audited last month. As a result, GAO recommended the attorney general to determine why the bureau failed to obey the disclosure requirements and to determine whether the software is correctly cross-referencing collected photos with images of criminal suspects.
The problem is that images of faces are protected personal data, according to the Privacy Act of 1974, which limits the collection, disclosure and use of personal data and requires agencies to disclose its kinds and ways of using it. However, the FBI did not publish such disclosures.
It is known that the FBI’s system searches not just its own database, but also other photo databases maintained by other agencies.
Initial queries about the program referred to the system that allows the bureau and some other agencies to cross-reference surveillance camera footage and other pictures with its collection of candidate photos. However, the recent report found a much larger program, run by the FBI, which conducts face recognition searches and is able to access external partners’ face recognition systems to support FBI active investigations.
Aside from privacy concerns, there is the question of the accuracy of the comparisons between photos of suspects and law-abiding citizens. Apparently, the results guaranteed 80-85% accuracy if the candidate is in the top 50 responses. In other words, it’s highly inaccurate. And as the database expands, the accuracy declines.
The state governments currently keep detailed facial recognition databases from driver’s licenses in order to stymie identity theft and other crimes, but the FBI uses this data for a very different purpose. In response, the FBI claimed it believed previous disclosures were adequate and its searches were in compliance with the US law.
Sourced from torrentfreak