Human-agent trust for government

FedRAMP, India Stack, eIDAS 2.0, and a thicket of national-security frameworks. Government AI demands the strictest human oversight — and is about to get the most regulatory pressure to prove it.

The unique constraints

Public-sector AI deployment differs from enterprise in three ways that matter for HATI:

• Citizen-facing accountability. Every decision must be traceable to a named, accountable public servant. "The system did it" is not a legal defense.
• Procurement frameworks. FedRAMP, FISMA, GovCloud — controls are prescriptive and audit-heavy. Vendors who build to the strictest controls expand to the broadest market.
• Sovereignty. Government data and workflows often cannot leave the country. Identity infrastructure must work without phoning home to a foreign vendor.

Three high-stakes flows

Benefits adjudication. AI agents that screen applications for unemployment, disability, healthcare, immigration. Article 14-equivalent oversight is mandatory; without HATI, benefit-denial decisions cannot be defended in administrative review.

Public safety. Predictive policing, AI-assisted threat assessment, automated traffic enforcement. Civil rights case law in the US and EU is converging on a "human in the loop with documented authority" requirement. HATI is the documented part.

Regulatory enforcement. Tax authorities, financial regulators, customs and immigration. Agent-driven case routing must show the human who authorized the routing logic, the human who reviewed the case, and the human who issued the action.

The architecture for sovereignty

Manav's open protocol means a sovereign deployment can run on national infrastructure. The protocol is not Manav-the-company's intellectual property to gatekeep. The reference implementation is open-source; relying parties verify cryptographically without trusting a US-hosted service. India Stack, eIDAS, and similar national identity systems can serve as Layer 1 anchors.

The procurement angle

FedRAMP-aligned vendors that document HATI primitives in their authorization-to-operate package will close federal sales 6–12 months faster than vendors who build proprietary identity. Similar timing dynamics in EU and India procurement. The lesson learned from the cloud era: open standards beat proprietary lock-in in government.

What we are seeing

Several US federal agencies have begun pilots with HATI primitives in regulated AI deployments. EU member states are aligning eIDAS 2.0 wallets with delegation chains for public-sector AI. India is exploring Aadhaar-anchored agent identity for benefit delivery. Each pilot is small; the trajectory is the same.

Common objections

The two pushbacks we hear from this vertical: integration risk — addressed by phased rollout starting with the audit trail (lowest risk, highest evidence-to-effort ratio), and internal politics — addressed by anchoring the project to a regulator deadline or a security-questionnaire deal-blocker, where the political question answers itself.

Frequently asked questions

What is the first integration to ship? The signed audit trail. It costs least, satisfies the most regulators, and produces the evidence everything else builds on. Every vertical we have integrated started here.

How does this affect end-customer experience? Invisibly, by design. The customer sees the same UI; the difference is in the audit log behind it. The latency added is single-digit milliseconds. The trust gain is structural.

What's the buying motion — security, compliance, or the line? Compliance writes the check; security signs off; the line of business sets the timeline. The strongest deals start with a regulator deadline; the next-strongest start with a deal-blocking security questionnaire.

Where to start

The first integration we recommend in this vertical: ai act article 14 playbook, then eidas 2 and manav. Both are deployable inside a quarter; both produce regulator-grade evidence; both unblock procurement conversations the rest of the stack depends on.

Adjacent reading

For the regulatory ground truth in this vertical, see the Article 14 playbook. For the integration shape, see audit-trail design. For the buying motion, see the CISO compliance stack. Most successful pilots in this vertical follow that order: regulation first, integration second, procurement third.

Where this goes first in the public sector

Three programs are likely to integrate before the rest. FedRAMP-authorized AI services. Already require continuous monitoring; the audit trail is a natural addition. India Stack agent flows. Aadhaar-anchored consent fits the Manav delegation shape almost directly. EU eIDAS-2 wallet integrations. Already specify selective-disclosure verifiable credentials; agent delegation extends the same primitives.

Procurement timelines are the limiting factor, not technology. The protocols are mature; the question is whether the agency can authorize a vendor on the schedule the program needs. The agencies that have started early are running 18-month head starts on the ones still scoping.

Why public-sector pilots become reference architecture

Public-sector pilots have the unusual property that successful implementations become reference architecture for entire jurisdictions. A federal agency that ships substrate-grade delegation infrastructure produces a procurement template that GSA-affiliated contractors inherit, a security architecture that other agencies cite, and an audit-trail format that becomes the de-facto standard for similar work. The leverage is enormous and underappreciated. We have invested disproportionate effort into public-sector pilots not because the contracts are larger — they are not — but because each successful pilot produces a public artifact that accelerates a hundred subsequent procurements. The strategy is the inverse of most enterprise sales: optimize for the artifact, not the contract. The artifact compounds. Vendors who optimize for the contract burn one cycle per deal. Vendors who optimize for the artifact bank reference architecture that compounds across years. We are playing the long version of this game on purpose, with explicit support from the public-sector buyers who understand the dynamics from their side.

Public trust requires private cryptographic accountability. There is no other path.