AI Safety and Geopolitical Policy Advocate works on public policies that reduce serious risks from advanced artificial intelligence while accounting for national security, international competition, economic interests, and democratic rights. The role combines AI governance, technical safety knowledge, geopolitical analysis, legal research, diplomacy, and public advocacy. Its purpose is to turn research about advanced AI risks into laws, standards, verification systems, corporate practices, and international agreements that decision-makers can apply.
This work is broader than asking companies to make AI systems safer. Advanced models depend on chips, data centers, energy, cloud services, skilled workers, supply chains, and international markets. Governments treat many of these resources as strategic assets. Policy advocates must therefore understand both the technical concerns surrounding advanced systems and the political incentives that shape how states and companies respond.
The field also requires practical judgment. Governments want economic growth, military readiness, technological independence, and influence over international standards. Companies want market access, investment, legal certainty, and room to develop products. Civil society groups want public protection, accountability, transparency, and respect for human rights. An effective advocate identifies policies that reduce serious risks while remaining politically workable, technically clear, and enforceable.
Why AI Safety Has Become a Geopolitical Policy Issue
AI safety has become a geopolitical policy issue because advanced models affect economic power, defense planning, cyber operations, information security, scientific research, and control over strategic infrastructure. States do not view AI only as a commercial technology. They increasingly treat compute capacity, semiconductor access, model development, cloud infrastructure, and technical talent as sources of national power.
This changes how safety proposals are received. A government that believes it is in a technological race will resist measures that appear to slow domestic development while rivals continue moving ahead. Even when two countries recognize a shared risk, they can disagree about the severity of that risk, the proper response, the information that should be disclosed, and the methods used to verify compliance.
International cooperation on AI safety therefore depends on the exact activity under discussion. Sharing sensitive model details can create security concerns. Joint work on verification methods, common testing language, incident categories, or shared technical protocols can offer safety benefits with lower strategic exposure. Research on cooperation between rivals identifies verification mechanisms and common protocols as promising areas because they can support oversight without requiring unrestricted access to sensitive systems.
An advocate must frame safety as part of national resilience rather than as an abstract restraint on innovation. Policies are easier to support when they protect infrastructure, reduce the chance of model theft, limit biological or cyber misuse, improve emergency readiness, and give domestic developers clear rules.
The Core Mission of the Advocate
The core mission of an AI Safety and Geopolitical Policy Advocate is to convert technical concerns into actions that governments, companies, international bodies, and the public can understand and implement. Research papers alone do not ensure that a government agency changes its rules or that a company improves its release process. Advocacy connects analysis with decision-making.
This work begins by identifying a specific decision. The decision can involve a proposed law, an agency rule, a procurement standard, an international declaration, an export-control measure, a model-testing requirement, or a company safety policy. The advocate then identifies who controls that decision, who influences those people, what incentives they face, and which policy outcome is achievable.
Professional advocacy often includes direct meetings with lawmakers, policy staff, regulators, corporate leaders, technical experts, journalists, and civil society groups. It can also include coalition building, public communication, legislative drafting, consultation responses, grassroots organizing, and support for whistleblower or incident-reporting protections. Effective advocates focus on the action they want a decision-maker to take and explain why that action serves the decision-maker’s own responsibilities.
The role does not require agreement with every participant on every AI issue. It requires finding areas of shared interest. A national security official can support stronger model-security standards because they reduce theft. A consumer group can support the same standards because they reduce harmful deployment. A company can support clear testing rules because predictable requirements reduce legal uncertainty.
Major AI Risks Covered by the Role
An AI Safety and Geopolitical Policy Advocate assesses risks that arise from model capabilities, deployment conditions, access controls, infrastructure dependencies, and deliberate misuse. The advocate does not need to conduct every technical test personally, but must understand the policy meaning of technical results.
Advanced AI systems can support cyber operations by helping users identify software weaknesses, generate malicious code, automate reconnaissance, or scale social engineering. Safety policy in this area can include secure development requirements, restricted access to sensitive capabilities, stronger identity controls, staged release procedures, and reporting duties for serious security failures.
Biological and chemical risks require close cooperation between model evaluators, life-science experts, security agencies, and public health specialists. Policy work can cover model testing for dangerous scientific assistance, access restrictions, monitoring of high-risk use, secure research environments, and rapid reporting when safeguards fail.
Autonomous systems create concerns about human control, escalation, targeting, accountability, and machine-speed decision-making. Advocates working on military or security policy must understand command responsibility, authorization rules, testing conditions, audit logs, and the limits placed on automated decisions.
Information operations present another major area. Generative systems can produce persuasive text, synthetic audio, fabricated video, impersonation attempts, and localized political content at scale. Policy responses include provenance systems, disclosure rules, platform duties, political-ad transparency, media authentication, and rapid response procedures during elections or security emergencies.
Model theft, unauthorized access, insider threats, and weak infrastructure security can allow powerful capabilities to spread without the original developer’s controls. Security requirements must cover model weights, training environments, cloud access, staff permissions, supply chains, and incident response.
Geopolitical Competition Changes What Is Politically Possible
Geopolitical competition shapes which AI safety policies governments consider acceptable. States often support safety in principle while rejecting measures that appear to give rivals an advantage or expose sensitive technical information.
Competition creates several recurring policy problems. Governments can disagree about what counts as a dangerous capability. They can distrust foreign testing laboratories. They can worry that inspections reveal trade secrets or national security information. They can also suspect that safety rules are being used to restrict market access or protect established companies.
Shared concern is therefore not enough. Durable cooperation requires a common understanding of the risk, a method for checking whether participants are following the agreement, and meaningful consequences when they do not. Analysis of geopolitical competition in AI argues that advocates must address these incentive problems directly instead of relying only on broad safety commitments.
An advocate should separate policies that require deep trust from policies that can operate under limited trust. A joint project involving sensitive model architecture requires strong controls. A shared format for reporting severe incidents requires far less technical disclosure. Common definitions, standardized tests, secure third-party evaluations, and crisis communication channels can produce value even when strategic rivalry remains intense.
This approach treats competition as a lasting condition. Safety policy must function even when diplomatic relations weaken, companies compete for market share, or governments change leadership.
Technical Cooperation Between Geopolitical Rivals
Technical cooperation between rivals works best when the safety benefit is clear, and the risk of transferring strategically sensitive knowledge is controlled. Verification systems and common protocols are strong starting points.
Verification mechanisms help one party confirm that another has followed agreed rules without receiving unrestricted access to sensitive models, source code, training data, or infrastructure. Possible methods include secure audits, controlled evaluations, tamper-resistant logging, independent testing, cryptographic verification, and limited disclosure of standardized results.
Common protocols create shared procedures for predictable situations. These can cover severe incident notification, model theft, dangerous capability discovery, unauthorized release, loss of control, major cyber exploitation, or cross-border misuse. A protocol can define the type of event, the reporting timeline, the responsible contact points, and the minimum information needed for a response.
Shared evaluation language can also reduce misunderstandings. Countries often use different definitions for frontier models, general-purpose systems, high-impact applications, systemic risk, and unacceptable use. Common measurement methods allow officials to compare results without forcing every country to adopt the same legal system.
Cooperation becomes more sensitive when it directly improves model capabilities, exposes national research priorities, reveals infrastructure weaknesses, or gives participants access to protected systems. Each proposal therefore needs a risk review covering capability transfer, sensitive information, misuse opportunities, reciprocity, access controls, and exit procedures.
Verification as the Center of International AI Governance
Verification is the process of confirming whether a developer, government, cloud provider, or other covered party is meeting an agreed safety requirement. It converts political promises into checkable conduct.
A strong verification framework begins with a precise obligation. A rule stating that a company must act responsibly is too vague to verify. A rule requiring a developer to complete specified evaluations before releasing a model above a defined capability threshold is clearer. The framework can then identify who performs the test, what data must be recorded, what results must be reported, and what happens after a failure.
Verification can operate at several levels. Internal assurance uses a developer’s own safety and security teams. External assessment uses approved independent evaluators. Regulatory review gives a public authority access to required records. Technical verification uses secure logs, hardware measurements, access records, cryptographic methods, or controlled testing environments.
Not every detail needs to become public. Governments can create confidential reporting channels for national security information and protected business material. Public reporting can focus on the existence of testing, broad risk categories, serious incidents, corrective action, and compliance status.
An advocate must also consider verification failure. A system can fail because the standard is vague, the evaluator lacks access, the developer supplies incomplete information, the regulator lacks expertise, or penalties are too weak. Effective policy anticipates these problems before an agreement is signed.
Building Risk-Tiered AI Regulation
Risk-tiered AI regulation applies stronger duties to systems that present greater potential harm. It avoids treating a low-risk writing assistant in the same way as a highly capable model used for cyber operations, defense analysis, medical research, or major public services.
A tier can be based on model capability, deployment context, number of users, access to sensitive tools, autonomy, compute used in development, or the seriousness of foreseeable harm. Different tiers can trigger different duties.
Lower-risk systems can be covered by basic transparency, privacy, consumer protection, and security rules. Higher-risk systems can require documented testing, independent assessment, access controls, incident reporting, human oversight, deployment monitoring, and clear responsibility within the developer.
Frontier systems can face additional duties. These can include pre-release capability testing, model-weight protection, secure development environments, documented safety cases, protected reporting channels for employees, emergency response plans, and notification of severe incidents.
Thresholds must be reviewable because model efficiency and technical methods change. A threshold based only on training compute can become less useful when developers produce stronger systems with fewer resources. Capability measures, deployment reach, tool access, and real-world performance provide additional signals.
The advocate’s task is to make each tier understandable. Developers need to know when duties apply. Regulators need testable criteria. Courts need language that supports consistent interpretation. The public needs enough transparency to judge whether the system is being supervised.
Corporate Policy Advocacy and Regulatory Incentives
Corporate policy advocacy influences how AI rules are written, delayed, narrowed, expanded, or enforced. Companies often publish proposals about safety, innovation, privacy, national competitiveness, open models, licensing, and international standards. These proposals should be examined alongside the company’s commercial interests.
A company can support regulation while disagreeing with its scope. It can support national rules but oppose state or regional rules. It can favor standards that it already has the staff and resources to meet. It can also oppose disclosure requirements that expose security weaknesses, internal decisions, or confidential business information.
High compliance costs can protect the public when they fund serious testing and security. They can also create barriers that established companies can absorb more easily than smaller developers. An advocate must distinguish necessary safety costs from requirements that mainly protect market position.
Public statements do not always reveal private policy activity. One source in the research set warns that collections of published company positions are incomplete and can present a selective picture. That warning supports a wider review of lobbying records, consultation submissions, legislative language, meeting disclosures, enforcement history, and changes between public and private positions.
The aim is not to treat every company position as dishonest. The aim is to identify incentives, test the policy reasoning, compare statements across time, and make conflicts of interest visible.
National Security and Economic Security Responsibilities
National security policy for AI covers more than military applications. It includes semiconductor access, cloud security, data-center resilience, electricity supply, telecommunications, research security, model protection, scientific misuse, and dependence on foreign providers.
Export controls can restrict access to advanced chips, manufacturing equipment, software, or technical services. These measures require careful design because they affect alliances, supply chains, domestic companies, foreign researchers, and the pace at which alternative technologies are developed.
Compute governance focuses on access to the large-scale computing resources used to train or operate advanced models. Policy options include reporting duties for very large training runs, customer verification for high-risk access, security standards for cloud providers, and special review of transactions linked to sanctioned actors.
Research security addresses theft, covert influence, undisclosed conflicts, and transfer of sensitive knowledge. Overbroad controls can damage legitimate science and international exchange. Narrow rules based on identifiable risks are easier to defend and enforce.
Economic security also includes resilience. Heavy dependence on a small number of foreign suppliers, cloud providers, chipmakers, or cable routes creates exposure to disruption and political pressure. An advocate connects AI safety with plans for redundancy, emergency response, trusted suppliers, and continuity of public services.
International Governance Without Unrealistic Expectations
International AI governance creates shared expectations, communication channels, technical standards, and response procedures across borders. It does not require every government to adopt identical laws.
Formal government talks can address treaties, bilateral agreements, export controls, security commitments, and official reporting systems. Expert-level dialogue can develop technical definitions, compare testing methods, study verification tools, and prepare proposals before political agreement is possible.
International summits can establish broad principles, but principles have limited value without implementation. An advocate should push for named responsibilities, timelines, measurable duties, review procedures, funding, and clear treatment of noncompliance.
Small agreements can be more workable than a single global settlement. Countries can begin with shared incident categories, joint research on verification, model-security standards, or emergency contact channels. Successful cooperation in one narrow area can support wider agreements later.
Regional differences also matter. Countries have different legal systems, technical resources, security priorities, economic goals, and levels of dependence on foreign AI services. Global rules should not assume that every government has the same enforcement capacity. Technical support, training, funding, and access to independent evaluation can help less-resourced countries take part on fairer terms.
Public Communication and Coalition Building
Public communication makes complex AI safety issues understandable without exaggerating them. Strong communication explains the system, the risk, the affected groups, the proposed response, and the limits of current knowledge.
Alarmist language can reduce trust when predictions are presented as certainty. Technical language can also exclude the people affected by policy. An advocate should use concrete scenarios, explain uncertainty, distinguish current harms from future risks, and avoid presenting every AI problem as an extinction-level threat.
Coalitions are strongest when participants support the same policy for different reasons. Worker groups can support protected reporting channels. Security experts can support model-access controls. Consumer groups can support transparency and accountability. Researchers can support access to testing information. Smaller companies can support clear and proportionate rules.
Advocacy can use insider and outsider methods. Insider work includes policy meetings, technical briefings, bill drafting, and regulatory consultations. Outsider work includes public education, media engagement, community organizing, and lawful campaigns. Research on AI safety advocacy stresses that poor advocacy can damage its own objective, while strategic communication and trusted relationships can improve access to decision-makers.
Skills Needed for the Role
An AI Safety and Geopolitical Policy Advocate needs technical literacy, policy judgment, legal reasoning, geopolitical awareness, communication ability, and relationship-building skills.
Technical literacy includes understanding model training, evaluations, fine-tuning, inference, tool use, autonomous behavior, access controls, model weights, compute, cybersecurity, and the limits of current safety tests. The role does not always require advanced programming, but it requires enough knowledge to work accurately with researchers and challenge vague technical statements.
Policy analysis includes comparing regulatory options, identifying affected parties, predicting implementation problems, reviewing costs, and writing recommendations. Legal skills help with statutory language, administrative procedure, privacy, liability, trade rules, national security powers, and international agreements.
Geopolitical analysis covers state interests, alliances, strategic rivalry, sanctions, export controls, technology supply chains, industrial policy, and diplomatic signaling. The advocate must recognize when a safety proposal will be interpreted as an economic restriction or security threat.
Communication includes policy briefs, speeches, consultation responses, media interviews, legislative testimony, public education, and direct meetings. Strong advocates study their audience and adjust the explanation without changing the underlying facts.
Publicly available career material in the source set also shows how legal training, AI governance education, policy research, summit participation, and work on digital sovereignty can contribute to this career path.
Career Routes Into AI Safety Policy Advocacy
Career routes into AI safety policy advocacy include government service, legislative staffing, law, political campaigns, public relations, journalism, technology policy, corporate affairs, cybersecurity, international relations, and technical AI safety research.
Government experience teaches how decisions are made, how agencies coordinate, how budgets affect implementation, and why policy language changes during approval. Legislative work builds skills in drafting, negotiation, hearings, stakeholder management, and political timing.
Legal work provides a base for regulation, rights analysis, compliance, contracts, liability, administrative law, and international rules. Communications and journalism build the ability to explain technical issues, work with media, identify public concerns, and produce clear material under time pressure.
Campaign experience teaches coalition management, audience analysis, volunteer operations, message testing, and rapid response. Technical research experience gives an advocate credibility when evaluating model tests, security controls, or scientific disagreements.
People entering the field can begin through policy fellowships, legislative placements, research assistance, public consultations, volunteer advocacy, technical courses, professional writing, or work on related issues such as privacy, cyber policy, consumer protection, election security, or digital rights. Career guidance in the source material recommends building relevant experience before seeking highly specialized advocacy roles because the field remains relatively small.
A Practical Advocacy Workflow
A practical advocacy workflow begins with a clearly defined policy decision. The advocate records the desired action, the decision-maker, the legal authority involved, the timing, and the conditions required for approval.
The next step is a stakeholder map. This identifies supporters, opponents, technical advisers, affected industries, civil society groups, government departments, media voices, and people trusted by the final decision-maker.
The advocate then prepares a policy package. It can include a short briefing, a detailed technical note, draft legal language, implementation steps, expected costs, risk controls, and responses to likely objections.
Message testing follows. National security audiences often respond to resilience, control, strategic advantage, and protection against misuse. Economic audiences focus on certainty, competitiveness, market access, and compliance costs. Rights-focused audiences examine accountability, discrimination, privacy, transparency, and access to remedies.
After engagement begins, the advocate tracks changes in language, support, opposition, implementation plans, and enforcement authority. The work continues after a law or commitment is announced. Rules without funding, staff, technical capacity, reporting systems, or penalties can remain ineffective.
Ethical Guardrails for Responsible Advocacy
Ethical advocacy presents risk accurately, discloses uncertainty, respects lawful processes, and avoids manipulation. AI safety policy affects civil liberties, scientific freedom, economic opportunity, national security, and the distribution of political power.
Advocates should separate established technical findings from forecasts and personal judgments. They should correct errors, disclose relevant funding relationships, protect confidential information, and avoid overstating consensus.
Safety policy should also include safeguards against abuse by governments. Surveillance powers, data access, licensing, export controls, and restrictions on research can be misused. Proposals should include legal limits, independent review, appeal rights, proportionality, and public accountability.
International advocacy requires respect for countries with different economic conditions and policy priorities. Rules designed by wealthy states can impose costs on countries that depend on imported models, cloud services, and technical infrastructure. Fair participation requires consultation, technical support, and attention to unequal bargaining power.
Lawful advocacy remains essential. Source material on advocacy warns that illegal or violent tactics damage public trust and the policy objective itself.
Measuring Policy and Advocacy Impact
Policy impact is measured through changes in decisions, conduct, implementation, and public understanding. Media attention alone does not show that risk has been reduced.
Useful indicators include adoption of clear safety duties, stronger testing requirements, protected incident reporting, improved model security, funding for enforcement, better coordination between agencies, and inclusion of verification procedures in international agreements.
Intermediate progress also matters. A technical term can enter draft legislation. An agency can request a new study. A company can revise its release policy. A coalition can agree on common language. A regulator can gain access to technical expertise.
Relationship quality is another indicator. Trusted access to policy staff, technical teams, and affected communities can improve future work. These relationships must not become uncritical. Access is valuable only when the advocate remains accurate, independent, and willing to challenge weak policy.
The advocate should document what changed, why it changed, who contributed, and whether the result is being enforced. This reduces the risk of attributing a broad political outcome to one campaign or organization.
The Future Scope of AI Safety and Geopolitical Advocacy
The future scope of this role will increasingly cover AI sovereignty, compute governance, semiconductor policy, model security, scientific misuse, autonomous systems, international verification, and the rights of countries that depend on foreign AI infrastructure.
Hardware and infrastructure policy will receive greater attention because control over chips, data centers, cloud access, energy, and communications affects which systems countries can build and supervise. Public discussion in the source material connects digital sovereignty with control over physical infrastructure and long-term governance options.
Policy advocates will also need to address the spread of capable open models, lower-cost training methods, agentic systems, and models connected to external tools. Safety rules based only on a small group of leading developers will become less complete as capabilities spread.
Verification will remain central. Governments will need methods that confirm compliance while protecting security and confidential information. Shared protocols will be especially useful for incident reporting, emergency coordination, and common technical definitions.
The strongest advocates will combine technical accuracy with political realism. They will recognize competition without treating cooperation as impossible. They will support innovation without treating every safety duty as an obstacle. They will defend public rights while responding seriously to national security threats.
AI Safety and Geopolitical Policy Advocacy is therefore a public-policy career focused on converting advanced AI risks into workable safeguards. Its value comes from understanding technology, incentives, law, security, and communication at the same time. The role succeeds when governments and companies move from broad promises to clear duties that can be tested, verified, enforced, and improved.
AI Safety and Geopolitical Policy Advocate helps governments, companies, researchers, and civil society respond to advanced AI risks through clear rules, technical standards, security measures, and international agreements. The role connects AI safety research with national security, economic policy, legal oversight, diplomacy, and public accountability.
Success in this field depends on more than identifying possible harm. Advocates must understand how governments compete, how companies influence regulation, how technical systems can be tested, and how compliance can be verified without exposing sensitive information. Effective policies need clear thresholds, defined responsibilities, reporting procedures, independent review, and consequences for noncompliance.
International cooperation remains possible even between strategic rivals when projects are narrow, verifiable, and designed to protect confidential information. Shared incident-reporting procedures, common evaluation terms, model-security standards, and emergency communication channels can reduce risk without requiring countries to abandon their security or economic interests.
This career suits people who can combine technical understanding with policy writing, legal reasoning, geopolitical analysis, and persuasive communication. As advanced AI becomes more connected to defense, cyber operations, scientific research, public services, and critical infrastructure, demand will grow for professionals who can turn broad safety commitments into practical and enforceable action.
AI Safety and Geopolitical Policy Advocacy: FAQs
What Is an AI Safety and Geopolitical Policy Advocate?
An AI Safety and Geopolitical Policy Advocate develops and supports policies that reduce risks from advanced artificial intelligence while considering national security, international competition, economic interests, and public rights.
What Does an AI Safety and Geopolitical Policy Advocate Do?
The role includes policy research, risk analysis, legislative drafting, stakeholder engagement, public communication, international cooperation, and the development of safety standards for advanced AI systems.
Why Is AI Safety Connected to Geopolitics?
AI affects military capability, cyber operations, scientific research, economic power, semiconductor access, cloud infrastructure, and information security. These factors make AI safety part of international competition and national strategy.
What Types of AI Risks Does This Role Address?
The role can cover cyber misuse, biological and chemical risks, model theft, autonomous systems, political manipulation, synthetic media, infrastructure vulnerabilities, and the uncontrolled release of highly capable models.
How Does an Advocate Influence AI Policy?
An advocate can prepare policy briefs, meet lawmakers, contribute to public consultations, draft regulatory language, build coalitions, support technical standards, and explain complex risks to decision-makers and the public.
What Is AI Verification in International Policy?
AI verification refers to methods used to confirm that governments, developers, cloud providers, or other parties are following agreed safety rules without exposing unnecessary confidential or security-sensitive information.
Why Are Verification Mechanisms Important?
Verification turns broad promises into actions that can be checked. It allows regulators and international partners to assess whether testing, reporting, access controls, and security duties are being followed.
Can Geopolitical Rivals Cooperate on AI Safety?
Yes. Cooperation is more practical when it focuses on narrow areas such as incident reporting, shared technical definitions, emergency communication channels, model-security standards, and controlled evaluation methods.
What Is Risk-Tiered AI Regulation?
Risk-tiered regulation applies stronger requirements to AI systems that have greater capabilities, wider reach, sensitive applications, or a higher potential for serious harm.
What Rules Can Apply to High-Risk AI Systems?
High-risk systems can face independent testing, access restrictions, incident reporting, human oversight, security reviews, deployment monitoring, documentation duties, and emergency response requirements.
How Do Export Controls Affect AI Safety Policy?
Export controls can limit access to advanced chips, manufacturing equipment, software, cloud services, and technical knowledge. They are used to address security concerns but can also affect trade, research, alliances, and supply chains.
What Is Compute Governance?
Compute governance covers policies related to the powerful computing resources used to train and operate advanced AI models. It can include reporting requirements, customer verification, security standards, and oversight of very large training runs.
How Do Companies Influence AI Regulation?
Companies influence policy through lobbying, consultation responses, public statements, technical proposals, trade groups, and direct engagement with governments. Their recommendations should be reviewed alongside their business interests.
What Skills Are Needed for This Career?
Useful skills include AI literacy, geopolitical analysis, policy writing, legal research, public speaking, stakeholder management, cybersecurity awareness, negotiation, and the ability to explain technical issues clearly.
Is a Technical Degree Required for This Role?
A technical degree is not always required, but a strong understanding of AI systems, model testing, cybersecurity, compute, access controls, and technical limitations is highly useful.
Which Academic Backgrounds Are Relevant?
Relevant backgrounds include public policy, political science, law, international relations, economics, cybersecurity, computer science, engineering, communications, and security studies.
Where Can an AI Safety Policy Advocate Work?
Possible employers include government departments, legislative offices, policy research groups, international bodies, technology companies, regulatory agencies, civil society organizations, and public-interest legal groups.
How Is Success Measured in AI Safety Advocacy?
Success can be measured through stronger regulations, funded enforcement, improved testing requirements, better incident reporting, clear international protocols, safer corporate practices, and increased technical capacity within government.
Why Will This Role Become More Important?
The role will become more important as advanced AI becomes more connected to defense, cyber operations, scientific research, elections, critical infrastructure, public services, and international competition.
