The National Reconnaissance Office (NRO), the agency that builds and operates U.S. intelligence satellites, frequently makes mistakes when it classifies national security information, according to an assessment performed last year by the NRO Inspector General.
“From the classified documents we reviewed at NRO headquarters, 114 of 134 documents contained classification errors,” the IG report said.
Agency classification officials “lack sufficient knowledge of classification principles and procedures necessary to perform their duties,” the NRO Inspector General found. “One OCA [original classification authority] had almost no knowledge of his responsibilities.”
“Because of the lack of full compliance in multiple areas, the NRO is susceptible to the risk of persistent misclassification,” the IG said.
The IG report was performed in response to the “Reducing Over-Classification Act of 2010,” which required the Inspectors General of all agencies that classify information to evaluate their classification programs. A copy of the report was obtained under the Freedom of Information Act by the GovernmentAttic.org web site.
Most of the classification errors discovered by the Inspector General are administrative rather than substantive. Like other IG evaluations conducted under the Reducing Over-Classification Act, the NRO Inspector General review does not allow for the possibility that an agency could be in full compliance with classification rules and nevertheless be overclassifying information.
Instead, the IGs have focused on errors in marking documents, failures to specify proper authorities or to cite responsible officials, and similar defects in conformity with established rules.
Still, these are not necessarily trivial failures. Between 2005 and 2012, for example, NRO improperly exempted records from automatic declassification at 25 years when it had no authority to do so, the IG said.
The Inspector General reviewed NRO classification guides (which dictate the classification levels of particular items of information) “and we found that all but one of the 62 guides had classification errors.”
Puzzlingly, the Inspector General also reported that NRO “has not conducted timely reviews [of] its security classification guides” and that “three of the 62 SCGs had not been reviewed within five years.”
This finding appears to be inconsistent with a 2012 NRO report which affirmed that all of its security classification guides — of which there were 67, not 62 — had been reviewed in response to the Fundamental Classification Guidance Review. An explanation of the inconsistency was not immediately available.
NRO officials “non-concurred” with the findings and conclusions of the Inspector General report.
The report contains “numerous sensationalized, exaggerated and misleading statements,” wrote A. Jamieson Burnett, the director of the NRO Office of Security and Counterintelligence.
Other previously disclosed IG reports issued in response to the Reducing Over-Classification Act addressed classification programs in the Department of Defense, Department of Justice, Department of Homeland Security, and the Environmental Protection Administration.
Perhaps the biggest incentive for reducing overclassification is the negative impact that unnecessary secrecy can have on government operations.
“A major impediment to operating with international partners is the U.S. tendency to classify information, complicating the crucial flow of important data to our allies as well as within and among our own Services,” according to a new article in Joint Force Quarterly, which is published by National Defense University for the Chairman of the Joint Chiefs of Staff.
“The U.S. military needs to […] try harder to communicate in the unclassified domain,” wrote Jeffrey M. Shaw in his article “Putting ‘A Cooperative Strategy for 21st Century Sea Power’ to Work,” Joint Force Quarterly, January 2014.
These ideas aim to advance the detailed policy solutions needed to foster public trust and implement fairness in the adoption of AI across diverse domains, from healthcare and government benefits to rural access, education, and worker protections.
The evidence is clear: algorithmic pay-setting is established in app-based work, and payroll/timekeeping failures show how software can produce systemic wage harm at scale
While a few states have taken steps to implement decision-making mechanisms for certain AI systems, too many leaders are simply accepting narratives about AI’s purported public benefit at face value – jumping to the “how” of AI implementation before thoroughly vetting potential systems and deciding whether they are appropriate to use at all.
When properly structured — with specific numeric targets, secured financial obligations, independent monitoring, and meaningful enforcement — CBAs transform data center deals into durable community partnerships.