Pentagon Must Produce Plan for Declassification
Updated below
The Department of Defense must explain by early next year how it is going to meet its obligations to declassify a growing backlog of classified records, Congress said this week.
A provision (sect. 1759) in the new House-Senate conference version of the FY2020 national defense authorization act requires the Pentagon to prepare a report including:
* a plan to achieve legally mandated historical declassification requirements and reduce backlogs;
* a plan to incorporate new technologies, such as artificial intelligence, that would increase productivity and reduce the cost of implementing such a plan;
* a detailed assessment of the declassified documents released in the past three years along with an estimate of how many will be released in the next three years;
* other policy and resource options for reducing backlogs of classified documents awaiting declassification.
While the new legislative language is a welcome acknowledgment of a persistent problem, it does not by itself significantly advance a solution. In particular, the legislation does not authorize any new funds for declassification or for development of new declassification technologies, which are not yet mature. Nor does it define an alternative in the event that DoD proves unable to meet its declassification obligations.
In a prior draft adopted by the House of Representatives, the CIA and the State Department would also have been required to prepare similar reports. But those requirements were dropped in the final bill.
“The U.S. government’s system for declassifying and processing historical records has reached a state of crisis,” wrote William Burr of the National Security Archive lately. See “Trapped in the Archives,” Foreign Affairs, November 29, 2019.
Update (October 2021): The required report on Reducing the Backlog in Legally Required Historical Declassification Obligations of the Department of Defense was released in October 2021.
Commercial artificial intelligence tools have recently emerged that are able to produce police reports. If the resulting reports are inaccurate, incomplete or biased, or if the process leaks confidential information, this could undermine the criminal justice system and harm citizens.
Too often, affected patients, clinicians, and regulators cannot see how the system works, why a decision was made, or whether meaningful human oversight occurred.
Existing tools from other domains, such as existing robust public engagement processes in drug development, when applied to AI deployment can help strengthen public trust in these systems and enhance perceptions of their legitimacy and the decisions they produce.
With thoughtful policy action, it is still possible to build systems that are fair, transparent, and accountable, and to earn the public trust that will ultimately determine AI’s future. We hope policymakers are ready to act.