Weekly Roundup - May 11-15, 2026

How Federal Agencies Can Build the Organizational Engine for AI at Scale. A May 12 FNN commentary argues that federal agencies have largely moved past the question of whether to adopt AI -- the Trump administration's AI Action Plan has made that decision for them. The harder, and largely unresolved, question is how. Successful AI implementation begins with the right framing: agencies that start with 'What is our AI strategy?' are likely to fail, while those that ask 'How can AI enable our strategy?' are positioned to succeed. The author identifies three organizational prerequisites for AI at scale -- strategic clarity, governance architecture, and an execution engine -- and argues that most agencies have the first but are missing the latter two. Technology is only as impactful as the policies and people behind it, and the gap between insight and execution is where most AI initiatives stall. The commentary offers a practical framework for senior leaders looking to move from AI aspiration to durable operational deployment.

AI Drives New Debate Around CISA Software Patching Deadlines. Growing concerns about AI-powered cyberattacks are fueling serious debate within the federal cybersecurity community about how quickly agencies must patch software vulnerabilities -- including whether standard patch windows should now be measured in days rather than weeks. The debate intensified following Anthropic's Claude Mythos Preview release, which demonstrated the capacity to autonomously identify and exploit software vulnerabilities. Trump administration officials reportedly considered compressing CISA's standard Known Exploited Vulnerabilities catalog patch deadline from weeks to three days. CISA has already moved in that direction: all four KEV catalog entries between May 6 and May 14 carried three-day deadlines, and the 2026 year-to-date average KEV deadline of 14.4 days is meaningfully shorter than last year's 19.7-day average. Former NSA Cybersecurity Director Rob Joyce noted the risk environment 'changed dramatically even before Mythos' due to large language models' ability to rapidly identify exploitable code. A federal CIO speaking anonymously said patching timelines 'have to get as close to immediate as possible,' but cautioned that blanket acceleration without prioritization could have the reverse effect, overwhelming security teams with deadlines that are impossible to meet uniformly across complex federal IT environments.

Anthropic, Code for America Pilot AI Tools for SNAP Eligibility Support. Anthropic and nonprofit Code for America have launched a pilot program deploying AI tools to support state-level SNAP eligibility workers as they navigate an increasingly complex federal rules environment. The effort is designed to help caseworkers interpret intricate federal policy guidance and manage the growing administrative demands associated with recent SNAP policy changes -- not to replace human judgment, but to reduce the cognitive burden on eligibility staff so they can focus on higher-value interactions with program participants. The pilot reflects a broader pattern emerging across social benefit administration: AI tools positioned not as decision-makers but as interpretation and compliance aids for workers operating under resource constraints. Code for America has positioned this work as a model for responsible AI deployment in sensitive, high-stakes government benefit contexts where accuracy and equity carry direct consequences for low-income households.

NIST Aims for Summer Release of AI Cybersecurity Guidelines. The National Institute of Standards and Technology is targeting a summer 2026 release for a new set of cybersecurity guidelines specifically addressing AI-driven threats across different types of emerging systems, including agentic AI and autonomous tools. NIST is developing draft guidance across several system categories as the federal government deepens its engagement in AI model risk assessments and as the Claude Mythos Preview episode has sharpened official interest in formal frameworks for managing frontier AI capabilities. NIST's Center for AI Standards and Innovation previously launched an AI Agent Standards Initiative in February 2026, and the summer guidelines will build on that work. The release comes as agencies navigate an accelerating AI deployment landscape with limited formal cybersecurity guidance specifically tailored to AI-specific threat vectors, attack surfaces, and governance requirements.

2026 Federal CIO Forecast: Shifting Priorities, Enterprise Focus. Recent MeriTalk reporting highlights a continued shift across federal agencies from AI experimentation toward operational deployment tied directly to mission delivery. Senior technology officials emphasized that agencies are increasingly prioritizing practical applications of AI that automate repetitive administrative functions, improve citizen services, and reduce operational burden on the workforce.

Public Sector AI Governance Speeding From Concept to Practice, Report Says. Federal cybersecurity leaders described growing interest in integrating AI tools into cyber defense operations to automate threat detection, accelerate incident response, and support vulnerability management. Officials noted that AI-enabled tools are increasingly viewed as essential for managing the scale and speed of modern cyber threats.

DOE Invests $94 Million in Nuclear Energy to Help Power AI Boom. The Department of Energy (DOE) awarded a collective $94 million in federal cost-shared funding to eight companies to support the near-term deployment of small modular reactors (SMRs). Announced Thursday, DOE said the advanced, light-water SMRs will bolster nuclear energy generation in the 2030s to expand U.S. power grid capacity to meet increased demand – driven in part by artificial intelligence (AI).

ITA Taps TBM to Guide IT Spending Decisions.  The Commerce Department’s International Trade Administration (ITA) is using technology business management (TBM) practices to build a clearer picture of how IT dollars are spent across the agency. Total cost of ownership (TCO) models, benchmarking, and budget execution analysis have become key tools for improving transparency and informing budget decisions, officials said.

OVERSIGHT, MANAGEMENT & ACCOUNTABILITY

Watchdog Recommends Nearly 100 Ways for Agencies to Save Tens of Billions. The Government Accountability Office's 2026 annual duplicative and overlapping federal programs report -- its most comprehensive since the series launched in 2011 -- identifies nearly 100 new actions agencies and Congress can take to eliminate duplication, reduce fragmentation, and achieve financial savings. GAO estimates that implementing prior recommendations in this series has already generated almost $775 billion in financial benefits. The 2026 report covers programs at the departments of Defense, IRS, and VA, among others, and includes opportunities tied to shared services, IT consolidation, and procurement reform -- all areas where AI-driven efficiencies could amplify savings if combined with structural changes. The report also targets programs where federal agencies have persistently failed to implement prior GAO recommendations, making the oversight body's role in accountability central to the current administration's stated efficiency agenda.

Federal Discipline Was Never Supposed to Be Punitive. The MSPB Appeal Framework Reflects That. A GovExec commentary by employment attorney Justin Schnitzer offers a pointed clarification of what the Merit Systems Protection Board's penalty review process is actually designed to do and what it is not. The MSPB's scrutiny of adverse employment actions is not focused on the severity of the misconduct; it is focused on whether the proposed penalty can fix the employee. That framing, corrective rather than punitive, reflects the civil service's foundational philosophy about the purpose of discipline in a merit-based workforce. The commentary arrives as hundreds of thousands of federal employees are navigating reductions, terminations, reassignments, and adverse actions, and as the Supreme Court weighs whether the president can remove MSPB members at will. Understanding the MSPB's actual legal standard matters acutely right now for agencies, employees, and the managers caught between administration directives and civil service protections.

Trump's Staggering Defense Budget Could Weaken Bipartisan NDAA Support. Sen. Jack Reed, the top Democrat on the Senate Armed Services Committee, warned May 13 that the Trump administration's proposed $1.5 trillion defense budget -- including $350 billion via a separate reconciliation package Republicans could pass without Democratic support -- is complicating efforts to maintain the broad bipartisan coalition that has historically made the National Defense Authorization Act a must-pass vehicle. Reed noted that some Democrats who have consistently voted for the NDAA may reconsider when the bill is seen as effectively endorsing a defense spending level paid for by cuts to social programs. Lawmakers are still waiting for detailed reconciliation language from the administration, having received only general topline figures. The management and budget implications are significant: the reconciliation package would dramatically expand Pentagon AI, space, and modernization accounts -- but only if it survives a political process that is more uncertain than the headline defense budget number suggests.

CYBERSECURITY & NATIONAL SECURITY

ODNI Assigns Two Officials to Lead Intelligence Coordination on Election Threats. The Office of the Director of National Intelligence has named two officials to lead the intelligence community's coordination on election threat assessment as the 2026 midterm election cycle accelerates. For months, the intelligence community's election threat posture had been unclear following the departure of the prior administration's election security leadership and the significant reduction in CISA's election security workforce. The two ODNI officials will be responsible for integrating threat intelligence from across the intelligence community and ensuring that relevant threat information reaches appropriate state and local election officials -- a function that CISA's now-shuttered election security office had previously performed at the operational level. Sen. Mark Warner had formally demanded answers from DHS Secretary Mullin in early May on how the government planned to protect elections without CISA's dedicated election security staff. The ODNI appointments represent a partial institutional response to that concern, though the structural gap created by CISA's diminished capacity remains unaddressed.

FedRAMP Director Details New Cybersecurity Service, 20x Expansion. The Federal Risk and Authorization Management Program (FedRAMP) is expanding its push to modernize federal cloud security with a new FedRAMP Cybersecurity Service designed to speed cloud authorizations while bringing rotating technical experts into the program. In an interview with MeriTalk, FedRAMP Director Pete Waterman said the initiative is part of the broader FedRAMP 20x effort to move the program toward automated, continuous security validation. At the center of the effort is the newly announced FedRAMP Cybersecurity Service (FRCS), which Waterman said will help FedRAMP scale technical expertise across agencies and industry.

AGENCY MODERNIZATION, DATA & INNOVATION

Federal Data Guide Aims to Demystify 'Patchwork' of Laws Governing Data Sharing. The University of California Berkeley's School of Information has released a Federal Data Field Guide designed to serve as a plain-language reference on the types of data the federal government collects, how those data are used, and what legal frameworks govern -- and limit -- their sharing. Former U.S. Chief Data Scientist Denice Ross described the guide as a public interest resource at a time when concerns about DOGE's access to sensitive federal datasets, the Social Security Administration's inspector general investigation into possible improper database access by former DOGE employees, and the GAO's findings on Treasury payment system security failures have all sharpened public attention to federal data governance. The guide arrives as agencies consider deploying AI tools that draw on federal data assets, raising practical questions about permissible use, data minimization, and the legal boundaries of sharing across agencies and with commercial AI vendors. The field guide explicitly aims to give the public and specialized stakeholders greater informational capacity to evaluate when their data may be at risk.

NSF Plans $1.5B Funding for ‘Breakthrough’ X-Labs Science Partnerships.  The National Science Foundation (NSF) announced on May 14 that it plans to invest up to $1.5 billion over the next decade in its existing NSF X-Labs initiative. The program is designed to support independent, milestone-driven research partnerships aimed at achieving what the agency calls “generational breakthrough science” outside traditional academic and corporate institutions. The funding plan announcement marks a major expansion of NSF X-Labs, an effort initially previewed in late 2025 under the working name Tech Labs. The initiative was created to foster new models of scientific research that bring together interdisciplinary teams of researchers, engineers, and entrepreneurs to solve specific scientific challenges. NSF describes the X-Labs model as independent organizations that compete for milestone-based federal funding.

WORKFORCE, LEADERSHIP & MANAGEMENT

State Dept Directs Managers to 'Revise and Recalibrate' Scores for Employee Evaluations They've Already Submitted. The State Department is directing some managers to go back and revise recently submitted annual performance evaluations to give employees lower scores -- a model that may soon extend across the federal government under new OPM guidelines targeting grade inflation. A State Department performance management reform email instructs HR officers to 'pull back EERs individually' if they were submitted before final score calibration, and strictly enforces an average rating cap of 3.25. Foreign Service officers told FNN that the direction to reconsider scores has been 'extremely common' -- and that all revisions are downward, not upward. Junior employees are being asked to accept lower ratings that remove them from promotion consideration, while more senior staff remain competitive. One FSO called the system 'extremely demotivating.' The State Department framing -- that a 3.0 (Fully Successful) is the appropriate score for excellent work, and that a 5 with weak justification is less competitive than a 3 with strong narrative -- reflects OPM's broader effort to enforce performance differentiation across a workforce accustomed to compressed high ratings. The model is significant for federal management more broadly: if State's calibration approach becomes the OPM-mandated standard, it will reshape how managers across all agencies communicate performance and how employees understand their standing.

Army Corps of Engineers Faces High Attrition Over Plans to Relocate NYC Office. The Army Corps of Engineers is drawing bipartisan congressional opposition and union resistance to plans to relocate hundreds of employees from federally owned office space in New York City to commercially leased space in New Jersey. The International Federation of Professional and Technical Engineers Local 98 said the agency began the search for new offices without consulting the union -- a violation of its collective bargaining agreement -- and that a lease is expected to be signed as early as May 18. A union survey found that 43 percent of affected New York District and North Atlantic Division employees living in New York City and Long Island would leave the Corps if their commutes were significantly extended by the move. House Republicans from New York, in an April letter, had warned the relocation 'could result in severe attrition in personnel' and called the loss of subject matter experts and community relationships 'a nonstarter.' The episode illustrates how real estate consolidation directives, when applied without workforce analysis or consultation, can produce talent losses that offset any cost savings -- a management dynamic playing out at multiple agencies across the government.

The Number of Feds in Tax Debt Spiked During the Pandemic. A new analysis finds that the number of federal employees carrying delinquent tax debt rose significantly during the pandemic years, with the delinquency rate growing from 4.9 percent in 2021 to 6.9 percent in 2024. The data arrive at a politically charged moment: the Trump administration has moved aggressively to use tax delinquency as a basis for federal employment actions, and Congress has previously debated legislation that would make tax delinquency grounds for dismissal of federal workers. The numbers put a quantitative frame on a phenomenon driven at least in part by the pandemic's broader financial disruption to the American workforce -- including federal employees -- rather than reflecting any unique pattern of noncompliance among government workers. The finding is relevant to ongoing workforce accountability debates and to the administration's use of financial conduct standards as part of its broader federal employment review process.

What Operating Rooms Can Teach Leaders About Team Design. A study of surgeries at a hospital in Madrid and a subsequent pilot program there offers lessons for any organization that relies on fluid, high-pressure teams. Hospitals have invested heavily in technologies to improve operating-room efficiency, yet performance variability persists. Research based on more than 77,000 surgeries in a Madrid hospital and a follow-up pilot program found that team design—not technology—is often the decisive factor. Teams perform better when members have prior experience working together, when staff rotation balances continuity with fresh exposure, and when team composition takes into account gender diversity. The pilot program significantly improved efficiency and reduced both readmissions and incidents in which a mistake or unsafe condition occurred during the surgical process but was detected in time to prevent injury to the patient.

Are You a Collaborative Leader? Wihout the right leadership, collaboration can go astray. Insead professors Ibarra and Hansen have examined what it takes to be a collaborative leader. They’ve found that it requires connecting people and ideas outside an organization to those inside it, leveraging diverse talent, modeling collaborative behavior at the top, and showing a strong hand to keep teams from getting mired in debate. In this article, they describe tactics that executives from Akamai, GE, Reckitt Benckiser, and other firms use in those four areas and how they foster high-performance collaborative cultures in their organizations.

Are You Meeting the Needs of the People You Lead?  Organizations often assume leadership succeeds or fails because of a leader’s style. But research on follower psychology suggests the bigger issue is alignment: Employees judge leaders based on whether they provide what people need most in a given moment—protection, fairness, vision, expertise, affiliation, or status. Drawing on research across the United States, the United Kingdom, and China, the authors argue that the best leaders are not defined by a single leadership style, but by their ability to diagnose shifting follower needs and adapt before misalignment erodes trust, engagement, and performance.

THIS WEEK @ THE CENTER 

RECENT BLOGS       

ICYMI – This week Michael J. Keegan welcomed Steven Hintze, Chief Data and Product Officer at the Arizona Department of Child Safety, to discuss what it takes to bring artificial intelligence into state child welfare systems, how you modernize legacy systems without losing the human judgment at the heart of child welfare, and what happens when you combine data, product thinking, and human-centered design to rethink how government delivers services.