The 2015 federal cyber law gave agencies a year to encrypt data and require MFA; a 2021 order set both again

The Cybersecurity Act of 2015, approved December 18 inside the omnibus, told each agency to encrypt sensitive data and require multi-factor authentication within a year, unless its head certified otherwise. In 2021 an executive order set both goals again, agency-wide, with a 180-day clock.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

The Cybersecurity Act of 2015, approved December 18 inside the omnibus, told each agency to encrypt sensitive data and require multi-factor authentication within a year, unless its head certified otherwise. In 2021 an executive order set both goals again, agency-wide, with a 180-day clock.

  1. The week On December 18, 2015, the Consolidated Appropriations Act, 2016 became Public Law 114-113. [1] The law’s text records it as approved December 18, 2015, and its legislative history lists the bill as H.R. 2029. [1] It carried the year’s appropriations in a series of divisions.
  2. The shared sensor: detection and prevention at the boundary The law asked for more than a fixed sensor.
  3. The internal defenses: finding intruders already inside A boundary sensor sees traffic entering and leaving.
  4. The agency’s own controls: data, encryption and authentication In our reading, the order of the list was a sequence.
  5. The paper trail: reports, reviews and a sunset The law demanded detail on the shared sensor.
  6. How the law reached different organizations For a large department with components, the law created several layers of work.
  7. The strongest objections The law set a deadline but no money or penalty in the provisions we reviewed. [1] A skeptic would say the duty was hortatory.
  8. What a program director would check today These are good practice drawn from the law and the orders that followed.
  9. Why the agency layer was the harder promise The shared sensor was a service one department could build and others could join.
  10. With hindsight GAO also reported on adoption.

The week

On December 18, 2015, the Consolidated Appropriations Act, 2016 became Public Law 114-113. [1] The law’s text records it as approved December 18, 2015, and its legislative history lists the bill as H.R. 2029. [1] It carried the year’s appropriations in a series of divisions. Division N could be cited as the Cybersecurity Act of 2015. [1]

Division N had four titles. [1] Title I was the Cybersecurity Information Sharing Act of 2015. [1] Title II, National Cybersecurity Advancement, had two subtitles: one on the National Cybersecurity and Communications Integration Center, and one titled the Federal Cybersecurity Enhancement Act of 2015. [1] Title III was the Federal Cybersecurity Workforce Assessment Act of 2015. [1] Title IV collected other cyber matters. [1]

Title I addressed sharing by non-federal entities. It authorized private entities to monitor their own information systems and to share and receive cyber threat indicators and defensive measures for cybersecurity purposes. [1] It required a non-federal entity, before sharing an indicator, to remove information it knows at the time of sharing to be personal information of a specific individual not directly related to a cybersecurity threat. [1]

Title I also told the Secretary of Homeland Security to develop, within 90 days, a capability and process to accept indicators and defensive measures from non-federal entities in real time. [1] It barred lawsuits against private entities for monitoring, and for sharing or receiving indicators, conducted in accordance with the title, subject to conditions. [1] It said the title creates no duty to share and no duty to warn or act on information received. [1] Its provisions were set to end on September 30, 2025, with an exception for actions taken before that date. [1]

The federal subtitle took a different approach. It addressed agencies’ own networks, and its duties were not voluntary. It amended the Homeland Security Act to add a section on a federal intrusion detection and prevention system. [1] The Secretary was to deploy, operate and maintain, within one year of enactment, a capability to detect cybersecurity risks in traffic to or from agency information systems. [1] A second capability was to prevent traffic associated with those risks, or modify it to remove the risk. [1]

The subtitle then turned to the agencies. Each agency head was to apply those capabilities to all traffic between an agency information system and any outside system. [1] The statute’s deadline read “not later than 1 year after the date of enactment of this Act or 2 months after the date on which the Secretary makes available” the capabilities. [1] It added “whichever is later.” [1] Agencies were to apply improved capabilities within six months of their availability. [1] The Department of Defense, national security systems and the intelligence community were excepted. [1]

A separate section, titled Federal Cybersecurity Requirements, set controls for each agency. [1] Under that section, “not later than 1 year after the date of the enactment of this Act,” each agency head was to identify sensitive and mission critical data and assess access controls to it. [1] The head was also to encrypt or otherwise render that data indecipherable to unauthorized users, whether stored or in transit. [1] The section also required a single sign-on trusted identity platform for agency public websites that require user authentication. [1] It required multi-factor authentication for remote access to an agency information system and for each user account with elevated privileges. [1]

That section allowed an exception. It did not apply to a system for which the agency head personally certified, with particularity, three things to the Director of OMB. [1] The head had to certify that implementation would be excessively burdensome, that the requirement was not necessary to secure the system and that the agency had taken all necessary steps to secure it. [1] The certification also had to go to the appropriate congressional committees and the agency’s authorizing committees. [1]

Title IV added inspector general reviews. Within 240 days of enactment, the inspector general of each agency operating a covered system was to report to Congress on the agency’s logical access controls and multi-factor authentication for privileged users. [1] The report was also to cover software inventories and the agency’s capabilities to detect data exfiltration. [1]

The Cybersecurity Act of 2015 opens with Title I, the information sharing law for companies. [1] For federal agencies, the more demanding text sat in Title II’s second subtitle. It told every agency outside the Department of Defense and the intelligence community to do things that a bank’s or a large health system’s security officer would, we think, recognize. Agencies were to find their sensitive data, encrypt it and put multi-factor authentication in front of remote and privileged access. [1] It gave them one year from enactment. [1]

Our argument is that the law split federal network security into a shared platform and an agency duty, and set them on different clocks. The shared platform, DHS’s intrusion detection and prevention system, got a deployment deadline and a sunset. [1] The agency duty got a one-year deadline and an exception any agency head could invoke by personal certification. [1] In 2021 an executive order set agency-wide multi-factor authentication and encryption as goals again, with 180 days to meet them. [3] In our reading, that restatement suggests the 2015 deadline had not closed the work it described.

We take the federal provisions in four parts:

  1. The shared sensor: DHS’s intrusion detection and prevention capabilities at the agency boundary.
  2. The internal defenses: tools and plans to find intruders already inside agency networks.
  3. The agency’s own controls: data identification, access review, encryption and multi-factor authentication.
  4. The paper trail: the reports, reviews and sunset dates that were meant to show whether the first three happened.

The parts do not overlap. The first is a service DHS runs, the second covers monitoring inside the network, the third is what each agency configures itself, and the fourth is how Congress would find out. After them we look at how the law reached different kinds of organization, take the objections and end with checks for a program director today.

The shared sensor: detection and prevention at the boundary

What the law required of DHS

The subtitle gave the Secretary of Homeland Security one year from enactment to deploy two capabilities for any agency to use, with or without reimbursement. [1] One would detect cybersecurity risks in traffic to or from an agency information system. [1] The other would prevent that traffic from passing, or modify it to remove the risk. [1]

The law asked for more than a fixed sensor. The Secretary was to regularly deploy new technologies and modify existing ones to improve the capabilities. [1] DHS was to assess, through operational test and evaluation, advanced technologies, including detection technologies beyond signature-based detection. [1] It was also to set up a pilot to acquire, test and deploy such technologies as rapidly as possible. [1]

What the law required of agencies

Each agency head was to apply the capabilities to all information traveling between an agency information system and any outside system. [1] The deadline was the later of the two dates the statute named: a year from enactment, or two months from DHS making the capabilities available. [1] When DHS improved them, agencies had six months to apply the improvements. [1]

In our reading, the phrase “all information” set a measurable standard. An agency that routed some of its internet traffic through the sensors and left other connections outside them would not meet it. In our reading, the requirement turned network architecture into a compliance question. Every new internet connection, cloud service or remote access path would need to be routed through the federal capability.

The law also settled a legal question. It said the Secretary may access, and an agency head may disclose to the Secretary, information traveling to or from an agency information system. [1] It said so notwithstanding any other provision of law that would otherwise restrict the disclosure. [1]

The law limited what DHS could do with that access. The Secretary could retain, use and disclose information obtained only to protect information and information systems from cybersecurity risks. [1] Activities had to be reasonably necessary for that purpose, and information retained no longer than reasonably necessary. [1] Users of agency systems had to receive notice that their communications could be accessed for protection. [1]

Private contractors helping DHS were bound too. They could not disclose agency traffic to anyone other than DHS or the agency that disclosed it. [1] They could not use it for any purpose other than protecting agency systems or administering the contract. [1] The law barred suits against such contractors for assistance provided in accordance with the section. [1]

Privacy review built into the program

The law put privacy review inside the program rather than beside it. DHS was to update the privacy impact assessment required by the E-Government Act periodically. [1] Within one year of enactment, the DHS Privacy Officer, consulting the Attorney General, was to review the program’s policies and guidelines for consistency with privacy laws. [1] That review was to cover the laws governing acquisition, interception, retention, use and disclosure of communications. [1]

For an agency privacy officer, those provisions set the questions to ask about any traffic inspection the agency relied on. What is collected, how long it is kept, who sees it and whether users have been told. The law answered them for the federal sensor. [1] An agency adding its own inspection tools would need its own answers.

Why the shared sensor mattered to agency programs

For a small agency without a security operations center, the sensor was a service it could not build alone. For a large department, it was one layer among several. The law set the same duty for both: route all external traffic through it. [1]

The design also shaped vendor relationships. An agency moving email or applications to a commercial cloud would need that traffic to reach the federal sensor, or would need DHS to extend the capability to the cloud provider. In our reading, the 2015 text did not address cloud traffic specifically. GAO raised that gap in a report published on January 28, 2016. [2]

The internal defenses: finding intruders already inside

Advanced network security tools

A boundary sensor sees traffic entering and leaving. It does not see an intruder moving between internal systems. Section 224 addressed that gap. [1] It told DHS to include advanced network security tools in its continuous diagnostics and mitigation work, to improve visibility of network activity and detect anomalous activity. [1] It allowed commercial and free or open source tools. [1]

OMB was to develop, and DHS to implement, a plan to ensure each agency uses such tools. [1] OMB and DHS were to review government-wide policies to prioritize network security monitoring within agency networks. [1]

Metrics that measured time

The same section changed what agencies would be measured on. DHS, with OMB, was to update FISMA security metrics to include measures of intrusion and incident detection and response times. [1] OMB was to increase public transparency on agency cybersecurity posture, including metrics for department components, small agencies and micro-agencies. [1]

In our reading, the metrics provision mattered as much as the tools provision. A metric based on time to detect asks a different question from a metric based on controls deployed. An agency could have a tool installed and still take a long time to see an intruder.

The intrusion assessment plan

The law also created a hunt function. It told DHS, in coordination with OMB, to develop and implement an intrusion assessment plan to proactively detect, identify and remove intruders in agency information systems on a routine basis. [1] OMB was to send the plan to the congressional committees within six months of enactment. [1]

The phrase “on a routine basis” set an expectation of repeated searches, not a one-time review. For a program director, the plan implied that an outside team might look inside the agency’s network on a schedule the agency did not set.

What an agency would have to build

The internal-defense provisions put work on agency teams even where DHS supplied tools. Someone had to place the sensors, keep them running and read the alerts. The law amended FISMA so that DHS’s assistance role covers operating and maintaining technologies as well as deploying them. [1] In our reading, that wording recognized that a tool installed and then left unmanaged adds little.

An agency also had to decide what to log and for how long. Detection depends on records of what happened before an alert. A shorter retention period saves storage and limits privacy exposure. A longer one gives investigators more to work with. The law did not set a period for agency logs.

The intrusion assessment plan raised a staffing question. Routine searches for intruders need people who know the agency’s systems well enough to spot what is unusual. A small agency might rely on DHS teams for that. A large department might build its own. Either way, the hunt needs access, credentials and a way to report findings to the system owner who must act on them.

The emergency directive

Title II also amended FISMA to give DHS a new tool. In response to a known or reasonably suspected threat, vulnerability or incident representing a substantial threat to an agency’s information security, the Secretary may issue an emergency directive to the agency head. [1] The directive could require any lawful action with respect to the system, including systems operated by another entity on the agency’s behalf. [1]

The law surrounded the power with procedure. DHS was to set thresholds and criteria, privacy and civil liberties protections and notice to affected third parties. [1] It was to adopt the least intrusive means possible and limit directives to the shortest period practicable. [1]

The agency’s own controls: data, encryption and authentication

Five duties with one deadline

Section 225(b) listed five duties for each agency head, all due “not later than 1 year after the date of the enactment of this Act.” [1]

  1. Identify sensitive and mission critical data stored by the agency, consistent with its information system inventories.
  2. Assess access controls to that data, the need for readily accessible storage and individuals’ need to access it.
  3. Encrypt or otherwise render that data indecipherable to unauthorized users, whether stored on or transiting agency systems.
  4. Implement a single sign-on trusted identity platform for agency public websites that require user authentication.
  5. Implement identity management, including multi-factor authentication, for remote access and for each user account with elevated privileges.

In our reading, the order of the list was a sequence. An agency could not encrypt sensitive data it had not identified. It could not judge who needed access without first knowing what the data was. The law tied the identification step to the inventories already required under federal information law. [1]

What each duty asked an agency to do

In practice, identification would mean deciding which data counted as sensitive or mission critical, and who owned it. [1] The law tied this to existing inventories of major information systems. [1] An agency had to go one level below the system, to the data the system held.

The access assessment asked three things: whether access controls fit the data, whether the data needed to be readily accessible and whether each person with access needed it. [1] In our reading, the third question is a periodic access review. It requires a list of who has access and a manager able to confirm each person still needs it.

Encryption, as we see it, required key management. Data encrypted with keys stored beside it protects little. An agency had to decide who held keys, how they were rotated and how encrypted records stayed retrievable for their full retention period. The 2021 order later qualified its own encryption goal by federal records laws, which in our reading reflects the same tension. [3]

The single sign-on duty faced the public. Each agency was to implement a single sign-on trusted identity platform, developed by the General Services Administration with DHS, for each of its public websites that requires user authentication. [1] In our reading, for a citizen that meant fewer separate accounts, and for an agency it meant relying on a shared service for its login page.

The multi-factor duty covered two groups: anyone with remote access and every account with elevated privileges. [1] The second group includes administrator accounts and, in practice, service accounts that scripts and applications use. Those accounts cannot always answer a second-factor prompt. An agency had to find them, decide which could use stronger methods and document the rest.

Why the one-year deadline was hard to meet

The deadline applied to every covered agency at once. [1] In our reading, three practical facts made it hard. The identification step asks about data, while the inventories the law cited are inventories of systems. [1] Encryption at rest depends on each system’s storage and application design. Multi-factor authentication for privileged accounts reaches into older systems that may not support it.

The law recognized the burden in one way: the certification exception. [1] It did not provide money for the work in the text we reviewed, and it did not allow a phased schedule by system type.

The certification exception

The exception required more than a waiver request. The agency head personally had to certify, with particularity, that the requirement would be excessively burdensome. [1] The head also had to certify that the requirement was not necessary to secure the system and that the agency had taken all necessary steps to secure it. [1] The certification went to the Director of OMB and to the appropriate congressional committees. [1]

In our reading, the second and third certifications were hard to make honestly for a system holding sensitive data. Saying encryption was “not necessary” to secure such a system would be a strong claim. The design pushed agencies toward compliance or toward a documented, signed, public-to-Congress risk acceptance.

Who the controls reached

The section applied to agency information systems. Its definition, borrowed from the amended Homeland Security Act, covered systems used or operated by an agency or by another entity on the agency’s behalf. [1] In our reading, contractor-operated systems therefore fell within the duty. The Department of Defense, national security systems and the intelligence community were excepted. [1]

The inspector general reports in Title IV touched the contractor link. They had to describe the agency’s policies for ensuring that contractors and other service providers implemented the software-inventory and exfiltration-monitoring practices the report covered. [1]

How the controls compared with private-sector practice

The five duties read like a list a bank’s or a health system’s security officer would recognize. In our reading, none of them was new as a control. What was new was a statutory deadline, a statutory exception route and a reporting line to Congress. [1]

Private organizations faced no equivalent duty in this Act. Title I’s sharing provisions were voluntary for them, and expressly created no duty to share. [1] A bank’s encryption and authentication duties came from its own regulators and contracts, not from this Act.

The paper trail: reports, reviews and a sunset

Reports on the sensor

The law demanded detail on the shared sensor. DHS was to report within six months, and annually after, on the status of the intrusion detection and prevention capabilities. [1] The report was to include privacy controls and the types of indicators used. [1] It was also to include the number of times the capabilities detected a risk and the number of times they blocked traffic. [1]

OMB was to report, within 18 months and annually, a list of each agency and the degree to which it had applied the capabilities. [1] The Federal Chief Information Officer was to assess the capabilities’ effectiveness, costs and benefits, including against commercial tools, within a set window. [1]

What the reporting cost and what it revealed

Each report in the law had to be unclassified, with a classified annex allowed. [1] That design balanced two needs. Congress wanted public accountability, and agencies did not want a public map of their weaknesses. An agency preparing its section of OMB’s list had to decide what could be said publicly and what belonged in the annex.

The reports also took staff time. Counts of detections, blocks, indicators and pilot technologies require collection systems that produce those numbers. [1] In our reading, the reporting provisions assumed a measurement capacity that the agencies and DHS would have to build alongside the controls.

Reports on the agency controls

OMB’s annual FISMA report was to include a list by agency of compliance with the section 225(b) requirements. [1] In our reading, that list was the enforcement mechanism. The law did not set penalties. It made each agency’s status visible to Congress.

Title IV’s inspector general reports added an independent view. [1] Each was to describe logical access policies, the controls and multi-factor authentication used for privileged users and, if not used, the reasons. [1] They were also to cover software inventories and capabilities to detect exfiltration, such as data loss prevention, forensics and digital rights management. [1]

What the reports could not show

Counts of detections and blocks show activity, not effect. A sensor that blocks many connections may be catching noise, and one that blocks few may be missing threats. The law asked the Federal Chief Information Officer to assess effectiveness, including against advanced persistent threats. [1] In our reading, that assessment was the only report designed to answer whether the sensor worked, rather than how busy it was.

The section 225 list had a similar limit. A list of agencies and their compliance status shows whether a control is in place. It does not show whether the data identified was the right data, or whether the multi-factor method resisted phishing. Those questions moved to later guidance. [4]

An outside assessment

The law also asked GAO to study, within three years, the effectiveness of the government’s approach to securing agency information systems. [1] That study was to include the intrusion detection and prevention capabilities and the intrusion assessment plan. [1]

Two sunsets

The sunsets were written in different forms. One was a calendar date; the other was a period counted from enactment, which later laws replaced with dates. [1] [5]

The law set two end dates. Title I’s information sharing provisions were to end on September 30, 2025. [1] The authority for the federal intrusion detection and prevention system, and the related reporting requirements, were to “terminate on the date that is 7 years after the date of enactment of this Act.” [1]

The agency control duties in section 225 had no sunset in the text. [1] In our reading, Congress treated the shared platform as an authority to revisit and the agency controls as a permanent expectation.

Encryption and inspection pull against each other

The four parts were designed together, but two of them pull in different directions. The shared sensor inspects traffic crossing the agency boundary. [1] The agency controls require encryption of sensitive data in transit. [1] Traffic that is encrypted end to end is harder for a boundary sensor to read.

The 2015 text did not resolve that tension. [1] GAO’s 2016 review noted that the sensor did not address malicious content in web traffic, a capability DHS then planned. [2] As agencies encrypted more traffic, a signature-based sensor at the edge would see less inside each connection.

In our reading, the practical answer was to move some inspection inward, to endpoints and applications, where data is decrypted for use. The 2015 law’s internal-defense provisions pointed that way. [1] The 2022 zero trust memorandum went further, saying federal applications cannot rely on network perimeter protections against unauthorized access. [4] For an agency architect, the lesson of the combination is that encryption plans and monitoring plans need to be drawn together.

How the law reached different organizations

A cabinet department

For a large department with components, the law created several layers of work. Each component’s external connections had to reach the shared sensor. [1] The department head had to identify sensitive data across components and either meet the controls or sign certifications. [1] The department’s inspector general had to report on privileged access and exfiltration controls within 240 days. [1]

The transparency provision had a specific reach here. It called for displaying metrics for department components to the greatest extent practicable. [1] In our reading, that made it harder for a department to present only an average.

A small agency or micro-agency

A small agency had the same duties with fewer staff. The shared sensor helped, since DHS was to make it available with or without reimbursement. [1] The agency controls did not get a smaller version. The transparency provision named small agencies and micro-agencies specifically. [1]

An agency’s human capital office

Title III reached outside the security office. Each agency head was to identify all positions requiring cybersecurity or other cyber-related functions and assign employment codes under the National Initiative for Cybersecurity Education. [1] The Director of the Office of Personnel Management, with NIST, was to develop the coding structure within 180 days of enactment. [1] Procedures to identify civilian cyber positions followed on a separate schedule. [1]

For an agency, the coding work was the base for any later claim about cyber staffing shortages. A position not coded could not be counted. In our reading, the workforce title gave the controls in Title II a staffing counterpart: an agency could not show who would do the work without first identifying the positions.

An inspector general’s office

Title IV gave each inspector general of a covered agency 240 days from enactment to report. [1] Covered systems included national security systems and federal systems that provide access to personally identifiable information. [1] For an inspector general’s audit staff, the report was a new, fixed-date assignment on top of the annual FISMA evaluation.

The law allowed some relief. The reports could rest in whole or in part on an existing audit or evaluation, and could be submitted as part of another report, including the annual FISMA evaluation. [1] An office that had already reviewed access controls could build on that work. One that had not faced a short schedule.

The report’s content set a standard for later audits. It defined multi-factor authentication as not fewer than two factors, such as something known, an access device provided to the user and a biometric characteristic. [1] It defined a privileged user as one with access to system control, monitoring or administrative functions. [1] Those definitions gave auditors a fixed reference when agencies described their own controls.

A federal contractor

A contractor operating a system on an agency’s behalf fell inside the definition of agency information system. [1] In our reading, the agency’s duties to encrypt and require multi-factor authentication flowed into the contract. The inspector general reports also had to describe how the agency ensured that contractors implemented its software-inventory and exfiltration-monitoring practices. [1]

A state, county or city

State and local governments were outside the federal subtitle. Its duties ran to federal agencies. [1] They could take part in Title I sharing as non-federal entities. [1] For a county IT director, the law offered liability protection for sharing done in accordance with Title I, and no new federal duty on its own networks. [1]

A bank or a health system

Private organizations met the Act through Title I. [1] For a bank, the sharing protections came with a carve-out for communications with its own regulator about a threat, which the DHS process did not replace. [1] For a health system, Title IV included a section on improving cybersecurity in the health care industry. [1]

The strongest objections

Objection one: a statute cannot make agencies encrypt

The law set a deadline but no money or penalty in the provisions we reviewed. [1] A skeptic would say the duty was hortatory.

The objection has weight. It misses the mechanism the law did use. Personal certification by the agency head, sent to Congress, made non-compliance a signed decision. [1] OMB’s agency-by-agency list made it a public one. [1] Those are weak tools compared with funding. They are not nothing.

The objection also overlooks who carried the duty. The law named the agency head, not the chief information officer, as the person who must act or certify. [1] In our reading, that choice moved the decision out of the IT budget line and onto the desk of the official who sets the agency’s priorities. A chief information officer asking for encryption money could point to a statutory duty that fell on the head of the agency personally.

Objection two: the shared sensor was the real protection, so agency controls mattered less

GAO’s review, published in January 2016, cut against this. It found the National Cybersecurity Protection System partially, but not fully, meeting its stated objectives. [2] It said the system compared traffic to known signatures but did not detect deviations from baselines of normal network behavior. [2] It said the system did not monitor several types of network traffic. [2]

A boundary sensor with those limits cannot replace encryption and strong authentication inside the agency. In our reading, the law’s design, with both layers, was sound. The question was whether the agency layer would be delivered.

Objection three: the 2021 restatement shows the law failed

The 2021 order did set multi-factor authentication and encryption as goals again. [3] That shows the work was not finished. It does not show the 2015 law had no effect. The documents cited here do not show each agency’s status in 2016 or 2021, so we do not claim a measure of progress either way.

What the restatement does show is a pattern. The 2015 law asked for a written certification from agencies that could not comply. [1] The 2021 order asked agencies unable to fully adopt the controls within 180 days for a written rationale. [3] In our reading, both tools were built for exceptions, which suggests both drafters expected some.

Objection four: sunsets weaken security programs

A sunset on the shared sensor’s authority could leave agencies unsure whether a service they depend on will continue. Planning for a multiyear security architecture around an authority that ends on a fixed date carries risk. Short extensions in appropriations laws, such as the September 2026 law that moved the end date from September 30 to December 11, 2026, add to that uncertainty. [5]

The case for the sunset is oversight. The law tied the sensor’s termination to the end of its reporting requirements. [1] In our reading, that meant Congress would revisit both together. We think an authority that lets DHS access agency traffic notwithstanding other law is the kind Congress may want to renew deliberately. [1] The cost is planning risk for agencies. The benefit is a recurring decision point on a broad access power.

What a program director would check today

These are good practice drawn from the law and the orders that followed. Requirements differ by agency and by the guidance in force, and agencies should check with counsel and OMB guidance on what applies.

  • Is the inventory of sensitive and mission-critical data linked to the system inventory, so encryption status can be reported system by system?
  • Does every remote access path and every privileged account require multi-factor authentication, and is each exception documented and signed?
  • Is all external traffic, including cloud traffic, routed through the detection services the agency relies on, and can the agency prove it?
  • Does the agency measure time to detect and time to respond, as the 2015 law asked FISMA metrics to include?
  • Do contracts for agency systems operated by others require the same data, encryption and authentication evidence?

Evidence an inspector general would ask for

An inspector general reviewing these controls would follow the 2015 law’s own order. The first request would be the data inventory and how it links to systems. [1] The second would be the list of privileged accounts and the authentication used for each. [1] The third would be software inventories and exfiltration detection capabilities. [1]

The fourth would be the exceptions. A signed certification or rationale for each system that lacks a control is evidence of a risk decision. A system without either is a finding.

Why the agency layer was the harder promise

The shared sensor was a service one department could build and others could join. The agency controls were work every agency had to do inside its own systems. [1] The 2015 law gave both a deadline. Only the shared platform got a sunset, and by 2026 Congress was extending it in appropriations laws. [1] [5]

The agency controls had no sunset and no extensions. [1] They were restated instead, in an executive order in 2021 and in OMB’s zero trust strategy in 2022. [3] [4] In our view, that difference reflects the nature of the work. An authority can be extended by a sentence in an appropriations law. Encryption and authentication across an agency’s systems have to be built.

With hindsight

GAO published its review of the National Cybersecurity Protection System on January 28, 2016. [2] It found the system partially, but not fully, meeting its stated objectives. [2] It said intrusion prevention was limited to the types of traffic monitored, and that the system did not address malicious content in web traffic, though DHS planned that capability in 2016. [2]

GAO also reported on adoption. It said the 23 agencies required to implement the intrusion detection capabilities had routed some traffic to NCPS sensors. [2] It said only 5 of the 23 were receiving intrusion prevention services. [2] It said agencies had not taken all technical steps needed, such as routing all traffic through NCPS sensors, partly because DHS had not provided network routing guidance. [2]

GAO found gaps in planning as well. It said DHS had not defined requirements to detect malware on agency internal networks or threats entering and exiting cloud service providers. [2] It said DHS’s performance metrics did not gauge the quality, accuracy or effectiveness of intrusion detection and prevention, so DHS was unable to describe the value NCPS provided. [2]

Executive Order 14028 was signed on May 12, 2021. [3] Section 3(d) said that within 180 days of the order, agencies shall adopt multi-factor authentication and encryption for data at rest and in transit. [3] It qualified the duty: to the maximum extent consistent with federal records laws and other applicable laws. [3] Agency heads were to report on progress every 60 days until full adoption. [3] Those unable to fully adopt both within 180 days were to provide a written rationale at the end of that period. [3]

OMB’s memorandum M-22-09, dated January 26, 2022, built on the order. [4] It required agencies to meet specific cybersecurity standards and objectives by the end of fiscal year 2024. [4] It said agencies must use strong multi-factor authentication throughout their enterprise, and that phishing-resistant MFA is required for agency staff, contractors and partners. [4] It set a goal that agencies encrypt all DNS requests and HTTP traffic within their environment. [4] We describe M-22-09 as issued and have not traced later changes to it here.

By 2026 the shared sensor’s authority was running on extensions. The 2015 law had said the authority would “terminate on the date that is 7 years after the date of enactment of this Act.” [1] On September 2, 2026, Public Law 119-103 amended section 227(a) of the Federal Cybersecurity Enhancement Act of 2015 by replacing “September 30, 2026” with “December 11, 2026.” [5] The same law made the same change to the Cybersecurity Information Sharing Act’s effective period. [5]

As checked on November 16, 2026, the later public laws we found on govinfo, through Public Law 119-138, did not change either date. Unless Congress acts again, both authorities are set to end on December 11, 2026. [5]

What held up from the 2015 text is its list of agency controls. Data identification, encryption and multi-factor authentication for remote and privileged access reappear, in stronger form, in the 2021 order and the 2022 memorandum. [1] [3] [4] The documents cited here do not show each agency’s status at the one-year deadline. They do show that later documents set new clocks for the same work. [3] [4]

Spotted an error? Tell us and we will check it and correct the article where needed.

  • Cybersecurity Act of 2015
  • Federal Cybersecurity Enhancement Act
  • EINSTEIN
  • multi-factor authentication
  • encryption
  • Executive Order 14028