Nvidia Part 4: Policies and Controls for Managing CEO Risk

A CEO should be able to explain how the company manages a conflict involving the CEO. The answer should identify who receives the disclosure, who makes the decision, what restrictions follow, and who checks compliance. A signed code of conduct is only the beginning of that answer.

Post 3 examined corporate influence over export policy. This installment turns to decisions within a company’s control: the conduct it prohibits, the information executives must disclose, and the authority directors give compliance professionals to enforce those requirements.

Nvidia’s independent directors announced on November 10, 2026, that they had retained outside counsel to conduct an independent investigation. That announcement was not a finding of wrongdoing. The allegations concerning espionage and a personal relationship’s influence on export policy remain premises for this discussion. The policy recommendations below should not be read as findings about Nvidia’s conduct or the effectiveness of its existing controls.

The governance objective is to identify conflicts, prevent misuse of authority and information, and protect people who raise concerns. Policies should define those business risks precisely enough that employees and directors know when to act.

Require timely disclosure of relationships and conflicts

A relationship policy should explain when a personal connection creates a business obligation to disclose. Relevant circumstances can include influence over hiring, compensation, procurement, investments, customer relationships, government engagement, or access to sensitive information. The company needs sufficient information to assess that connection while limiting unnecessary intrusion into private life.

Nvidia’s public Code of Conduct identifies personal relationships among potential conflict concerns, directs employees to Employment Legal about perceived or actual conflicts, and references a conflict preapproval process. Those provisions establish a public policy framework; they do not establish what was disclosed or how a particular situation was handled.

Reuters’ analysis identified disclosure as an important governance question while noting that a code violation had not been established. That distinction matters: a relationship, a conflict, a disclosure failure, and a policy violation require separate factual assessments.

A workable policy should require disclosure before the employee participates in an affected decision, followed by prompt updates when circumstances change. Annual certifications can reinforce this obligation, but a company should not wait until the next questionnaire to learn about a developing conflict.

For the CEO, the reporting route should lead to a designated independent director or committee. The CEO should not decide whether the CEO’s own relationship requires review. The policy should expressly prohibit knowingly concealing a covered conflict, providing materially incomplete information, or using another employee to circumvent disclosure.

Internal disclosure and public securities disclosure require separate analysis. Item 406 of Regulation S-K addresses codes promoting ethical handling of actual or apparent conflicts between personal and professional relationships. Item 404 addresses specified related-person transactions. A company should not limit its internal conflict review to matters that meet public transaction-disclosure requirements. The failure this policy addresses is an undisclosed personal interest influencing a business decision before an authorized reviewer can evaluate it.

Make recusal effective in the actual decision process

Disclosure starts the review. The company then needs a decision about whether the conflict can be managed and, if so, how. Nvidia’s Code of Conduct calls for immediate withdrawal from a decision-making role when conflicts arise. For any company, implementing that principle requires attention to both formal authority and informal influence.

Consider a hypothetical CEO whose personal relationship creates a conflict in a supplier selection. Leaving the approval meeting accomplishes little if the CEO has already instructed the procurement team which supplier to select. A recusal plan should address recommendations, instructions to subordinates, negotiation, access to relevant deliberations, and subsequent attempts to reverse the decision.

Document who takes over, what authority that person receives, which communications are restricted, and when the arrangement will be reviewed. Change approval routing where necessary. Give the replacement decision-maker a direct escalation route if the recused executive intervenes. Some conflicts cannot be managed through recusal alone. Independent directors should have authority to require a change in responsibilities or discontinue the affected arrangement. The executive’s commercial importance should not determine whether the control applies. The failure this policy addresses is continued influence by a conflicted executive after a nominal disclosure or withdrawal.

Match foreign contact reporting to the person and the access

Foreign-contact policies require precision. A company must distinguish government security obligations applicable to covered personnel from controls it adopts to protect commercially sensitive information.

For covered personnel in cleared industry, the Defense Counterintelligence and Security Agency’s guidance on Security Executive Agent Directive 3 describes reporting of specified contacts, including unofficial contact with a known or suspected foreign intelligence entity and certain continuing relationships with foreign nationals. The guidance also states that limited or casual public contact is not reportable absent another applicable reporting requirement. These obligations do not automatically extend to every employee who handles valuable corporate information.

A corporate policy for other sensitive roles should define its own risk-based triggers, consistent with applicable privacy, employment, and nondiscrimination law. Those triggers might include requests for restricted technical information, attempts to bypass approved communications, threats or coercion, undisclosed payments connected to access, or a relationship creating a conflict over information-sharing decisions.

An employee should be able to report suspicious conduct without first proving that someone is an intelligence operative. Equally, nationality or ethnicity alone should not be treated as evidence of improper conduct. Reporting enables assessment; it does not establish wrongdoing. Assign responsibility for that assessment among security, compliance, legal, and export-control personnel. Specify what must be reported, to whom, within what period, and how sensitive personal information will be protected.

Reporting also needs an information-access control behind it. NIST’s least-privilege control, AC-6, provides a useful design principle: authorize the access needed for assigned tasks. It is a control framework, not a blanket legal requirement for every company. Apply that principle through access reviews, approved sharing channels, and proportionate logging. Seniority should not automatically authorize unrestricted copying or disclosure of sensitive information. The failure these policies address is an unassessed contact or pressure point enabling unauthorized access, disclosure, or interference with business decisions.

Protect reporting and participation in investigations

A reporting system must work when the subject of the concern controls careers, budgets, and access to the board. Provide more than one reporting route, including a confidential or anonymous channel where lawful and direct escalation to an appropriate independent director for concerns involving the CEO. Configure case routing so a complaint is not automatically delivered to its subject. Establish an alternative when the normal recipient has a conflict.

Mary Shirley’s commentary provided for this series makes a particularly useful point: protection should extend to people cooperating with compliance processes and to compliance professionals doing their jobs in good faith. That principle should appear in the policy and in the procedures used after a report.

Nvidia’s public code already prohibits retaliation against people who report concerns in good faith or cooperate with an investigation, including when allegations are unsubstantiated. The operational question for any company is how it identifies and responds to retaliation.

Consider changes in assignments, compensation, reporting lines, meeting access, and performance assessments. Establish an independent review process for potentially retaliatory actions. Protect confidentiality as far as circumstances permit without promising secrecy the company cannot maintain. Prohibit witness intimidation, interference with interviews, and destruction or alteration of relevant evidence.

Internal reporting policies must also preserve lawful external reporting. SEC Rule 21F-17 prohibits actions that impede direct communications with the SEC about possible securities-law violations. Review investigation instructions, confidentiality agreements, training, and separation documents together so one document does not undermine another.

The SEC’s settled administrative action, In the Matter of KBR, Inc., illustrates the risk. The SEC found that an investigation confidentiality statement requiring legal-department authorization for outside discussion violated Rule 21F-17, even though it was unaware of an instance in which an employee had been prevented from communicating with SEC staff. KBR settled without admitting or denying the findings.

Preserve legitimate confidentiality and privilege while ensuring that instructions do not impede protected reporting. Employees should not have to obtain the subject executive’s permission to raise a concern. The failure these policies address is the suppression of information through fear, retaliation, or restrictive procedures.

Give independent directors authority over executive exceptions

Executive oversight should be designed before a concern arises. The board should identify who receives CEO-related reports, who can authorize an investigation, who approves interim measures, and who assesses discipline or other corrective action. The CCO needs access to that decision-making structure. The ECCP asks about compliance autonomy, resources, access to relevant information, and direct access to the board or audit committee. These are useful tests of whether compliance can function when senior management is implicated.

Translate those principles into private board sessions, defined escalation rights, and independent review of material changes to the CCO’s role or resources during a sensitive matter. Where appropriate, employment terms can support those protections. A contract alone cannot supply authority that the board is unwilling to exercise.

Exceptions also require discipline. Nvidia’s code requires written Compliance Committee approval for waivers and disclosure of director or executive-officer waivers as applicable rules require. For any company, the review should establish that the approver is free of the relevant conflict and that the decision, rationale, conditions, and review date are recorded.

A managed conflict and a waiver of a code provision are different decisions. Counsel should assess the applicable approval and disclosure requirements. A policy should prohibit executives from approving their own exceptions or using subordinates to provide approval on their behalf.

Consistent accountability also matters. The US Sentencing Guidelines call for governing-authority oversight, adequate compliance authority and resources, and consistent enforcement through incentives and discipline. Executive status should not remove a person from that framework. The failure this policy addresses is executive control over the process intended to evaluate and constrain executive conduct.

Test the controls against the failures they address

The board should ask management to demonstrate what happens when these policies are used. DOJ’s compliance guidance specifically examines control testing, reporting of results, and tracking of corrective actions.

The following tests translate the policy objectives into evidence directors can review.

Policy Failure it is designed to prevent Evidence that the control operates
Relationship and conflict disclosure An undisclosed interest affects a company decision Timely disclosures, independent assessments, and updated mitigation plans
Recusal A conflicted executive continues directing the outcome Changed approval routing, a designated replacement, and review of interventions
Sensitive-contact reporting and access controls Unassessed pressure or contact leads to unauthorized information sharing Documented triage, access reviews, and resolved exceptions
Speak-up and non-retaliation Fear or interference suppresses reports and cooperation Tested escalation routes, follow-up with participants, and review of potential retaliation
Executive oversight and waiver approval The subject executive controls the review or grants an exception Independent approval records, private CCO access, and tracked corrective actions

Run a tabletop exercise using a hypothetical CEO conflict. Trace the report from intake through independent assessment, recusal, information protection, investigation, and follow-up. Test whether the process still works when the CEO disputes the concern or the designated director has a conflicting relationship.

Before closing the exercise, directors and the CCO should be able to answer four questions:

  1. Who can act on a concern involving the CEO without first obtaining the CEO’s approval?
  2. Which controls would actually change the executive’s access or decision-making authority?
  3. Who protects reporters, witnesses, and investigators from retaliation?
  4. What evidence will show the board that corrective actions remain effective?

Assign an owner and deadline to every gap. The value of the exercise lies in correcting the weaknesses it exposes.

These policies give a company a practical way to identify executive conflicts, protect sensitive information, and act when concerns reach the top of the organization. Their effectiveness ultimately depends on directors and compliance leaders being willing and able to use them. Tomorrow in our final blog post in this series, we bring it all together by examining what corporate America can apply across industries and what remains specific to Nvidia’s technology, leadership structure, and geopolitical exposure. The focus moves from individual controls to the broader governance system: independent oversight, protection of the compliance function, succession planning, consistent accountability, and evidence that the program works when its most powerful executive is involved.

Leave a Reply

Your email address will not be published. Required fields are marked *

What are you looking for?