Key Takeaways from Mayer Brown’s Defense and Security Forum

On October 13, 2026, Mayer Brown hosted its Defense and Security Forum in its Washington, DC office. The event was led by partners Raj De, who leads the firm’s Cybersecurity & Data Privacy and National Security practices, and Adam Hickey, a partner in the firm’s Cybersecurity & Data Privacy, National Security, and Global Investigations & White Collar Defense practices.

The forum brought together in-house counsel, compliance leaders, and government contracts and security professionals from companies across the defense, aerospace, and adjacent technology sectors. The program was designed to foster candid, peer-to-peer dialogue on the compliance challenges facing global organizations and to let participants exchange practical approaches to rapidly evolving regulatory, enforcement, and business risks.

Across four sessions, participants discussed the following topics:

• Government contracts, including the FAR rewrite, other transaction agreements (OTAs), CMMC, and supply chain risk;  
• Foreign regulatory considerations, including the EU and UK cyber and AI regimes;
• U.S. trade controls, including export controls, chip diffusion, and inbound (CFIUS) and outbound investment review; and artificial intelligence (AI) governance, security; and
• Agentic AI security and governance.

Below, we summarize key discussion points and takeaways from each session.

Government Contracts: FAR Rewrite, Other Transaction Agreements (OTAs), CMMC, and Supply Chain Risk

The opening session, led by partners J. Ryan Frazee and John Prairie, addressed the FAR overhaul, OTAs, CMMC, and supply chain risk.

On the FAR overhaul, the discussion began with the interplay between the interim rules already in effect and the proposed rules still subject to notice and comment. Participants noted that the proposed rules largely track the deviations agencies have already adopted, though with notable differences, leaving companies working against more than one framework at once. Much of the conversation concerned what the overhaul leaves unclear or does not address at all, and the difficulty of advising a business when the governing text may still change. Attendees observed that the new clause language has not yet driven materially different outcomes, though the potential is there, and that GAO has begun requiring agencies to identify the operative framework in their solicitations.

Turning to OTAs, participants discussed how far the instrument now reaches beyond prototypes in the conventional sense, extending to novel applications of commercial technology and, increasingly, to AI. That breadth prompted discussion of the effect on competition, and whether moving requirements onto OTAs narrows the field or simply relocates the competition into different procedures.

The group was direct about the risks of taking an OTA on boilerplate terms. The first concern was the statutory basis for clauses that appear in these agreements, and whether a clause carried over from a standard template actually rests on authority that applies to the transaction at hand. That question runs directly into the Christian doctrine, and the two points met there, because provisions grounded in statute are the ones most likely to be read into an agreement, while the doctrine’s application has otherwise narrowed since its original articulation, so a clause the government simply omitted is unlikely to be supplied after the fact. The written terms are therefore largely the deal, which limits the prospect of unbargained-for terms being imposed later but which also means a company cannot count on any background correction for a clause set it should have examined at the outset. That is what makes strategies for folding OTAs into an existing procurement process worth attention, so the terms receive the same front-end diligence as a traditional award and the instrument’s speed does not displace that review. Participants also noted that disputes under these agreements often arise on thin terms and without the familiar remedial path, because OTAs are not subject to the Contract Disputes Act.

On CMMC, participants discussed what may have caused the pause in the Phases 2 and 3 implementation, and whether this moment truly reflects a pause, as opposed to a wholesale change in direction. Participants also focused on when the proposed rule establishing a framework for safeguarding Controlled Unclassified Information (CUI) and reporting CUI incidents to civilian agencies (discussed in this Legal Update) would be effective, and the scope of the federal government’s characterization of information as CUI (as opposed to Federal Contract Information), which attendees described as a recurring pressure point.

A related discussion concerned certified third-party assessment organizations, including how assessments already in place will be treated and whether completed work retains its value. Because such assessments are not currently required across the board, prime contractors are absorbing more of the verification work themselves, along with incremental increases in their own exposure. That prompted concern about the subcontract layer, where a prime remains answerable for the posture of suppliers whose assessments it cannot compel.

Participants also discussed whether enabling small businesses forms part of a broader administration priority of removing barriers to participation in the bidding process. That opened a wider exchange about the policy goal of opening federal contracting to more participants. Entry may not have become appreciably easier in practice, and in some respects the compliance expectations placed on new and smaller entrants have moved in the opposite direction. The group discussed how best to communicate that frustration to the administration, including which channels have proven useful, when industry is better served by speaking collectively than individually, and how to frame the issue around specific barriers rather than general dissatisfaction.

The session closed on supply chain risk. The Department of War’s Anthropic ban drew attention, with attendees focused on next steps in light of the current legal posture. Implementation by agencies and primes has been uneven and the governing authorities remain unsettled. The conversation centered on how far the restriction extends and how to distinguish use of software in performance of a contract from other uses. Participants observed that the ban may bear on procurement, integration, and continued reliance in addition to direct use.

Foreign Regulatory Considerations: EU AI Act, Cyber Resilience Act (CRA), NIS2, and UK Considerations

The second session, led by partners Ana Bruder and Oliver Yaros, covered the EU AI Act, the Cyber Resilience Act, NIS2, and United Kingdom considerations, concentrating on the European cyber regime and the position of global companies subject to several frameworks at once.

The discussion centered on audit requirements and the scoping of NIS2. Participants described scoping as the threshold problem, since the operative question is whether in-scope activities are carried out in the EU and, if so, by which entity, so the analysis turns on corporate structure rather than headquarters location. Attendees worked through the situation in which an entity would not be captured on its subsidiary relationships alone but is nonetheless drawn into scope as part of a global group carrying out relevant activities in the EU.

Further complicating the analysis, each member state implements the directive in its own way. The managed service provider issue was the clearest illustration, because the category is newly covered, member states have implemented it differently, and the question of which entities it actually reaches therefore has different answers in different jurisdictions.

Participants exchanged strategies for global companies attempting to comply with many regimes at once, with considerable interest in standardizing cyber and reporting obligations. The appeal of standardizing to the most demanding applicable standard is that one set of controls and one reporting process can satisfy several regulators. The difficulty is that divergent national implementation can defeat that approach precisely where obligations differ in substance rather than degree.

A candid portion of the discussion addressed where EU cyber regulators are, and are not, communicating with one another. Attendees described the practical consequence of inconsistent expectations when the same incident must be reported in several jurisdictions on different timelines. That led to a discussion of how and when to engage with Brussels, with the prevailing view that engagement is most effective before positions harden into published guidance, and that its timing often matters more than its substance.

Trade Controls: ITAR/EAR Export Controls, CFIUS, AI Diffusion Rule, ICTS, and Outbound Investment

The third session, led by partners Aiysha Hussain and Tim Keeler, addressed ITAR and EAR export controls, the regulation of advanced chip diffusion, information and communications technology and services (ICTS), CFIUS and outbound investment.

Participants began with the logic of allowing non-state-of-the-art chips to go to China. The discussion examined whether permitting trailing-edge capability while restricting the frontier is a coherent dividing line, and the business planning difficulty that arises when that line can be redrawn quickly. Attendees noted that apparent permissiveness here sits alongside a heavier diligence burden in practice.

The group then discussed the appetite and strategy for deregulation, including the role of the technical advisory committees in identifying controls that no longer serve a purpose. A related exchange concerned which items on the Commerce critical list genuinely belong there given advances in technology. The concern was that items can remain listed well past the point of utility while genuinely sensitive capability moves faster than the list can track, so the list imposes cost without a corresponding security benefit.

Finally, participants discussed the extent to which companies can generalize across Five Eyes countries and more problematic countries alike for supply chain and CFIUS risk purposes. The practical question is whether a company may treat allied jurisdictions as a single tier in its risk assessments, or must assess each separately. Attendees noted that allied coordination is real in some areas while bilateral arrangements are negotiated separately in others, which cuts against treating any group of allies as interchangeable for diligence purposes.

Artificial Intelligence: Governance, Security, and Agentic AI Considerations

The forum’s final session, led by partners Stephen Lilley and Adam Hickey, concerned agentic AI governance and security, with recent publicly reported incidents as the backdrop.

Against that backdrop, participants discussed building low-risk off-ramps for certain use cases so that review capacity can be concentrated where it matters. The companion question was how often to check in on deployments once approved, since a use case cleared as low risk may change over time. The discussion focused on ways lawyers can and should help shape technology deployment decisions, given the potential for legal risk.

The group discussed appropriate levels of human oversight, particularly in the context of decentralized organizations where business units adopt tools independently. Other topics included the relative risks of internally built agents compared to externally procured ones, as well as shadow AI.

Liability questions drew particular attention. Attendees also discussed the SAFETY Act, and how its protections may apply in this context.

The session closed on strategies for keeping the board informed, with participants exchanging views on reporting cadence, which metrics are meaningful, and how to describe risk to a board asked to approve continued deployment.