TL;DR
- Adviser Exit: Sriram Krishnan, outgoing federal AI adviser, will leave his formal government job in June 2026.
- Cyber Reviews: Federal officials are advancing voluntary model access, cyber benchmarking, and a clearinghouse for vulnerability and patch coordination.
- Procurement Shift: National-security agencies have 120 days to update processes for onboarding advanced AI models from multiple suppliers.
- Open Caveat: Available material does not establish a reason for the departure, and Krishnan may continue as an outside adviser.
In June 2026, Sriram Krishnan, the outgoing federal artificial intelligence adviser, will leave his formal government job. Federal cyber-testing and national-security supplier deadlines are moving ahead as he exits.
Recent federal actions now affect AI developers, cyber agencies, key-infrastructure operators, and national-security procurement teams. Available material does not establish a verified reason for Krishnan’s departure.
Krishnan’s next phase would involve large AI challenges for the United States and its allies.
🇺🇸🚀 SOME NEWS: I’ll be leaving my role at the White House at the end of this month. After a break I’ll be working on helping tackle some of the large challenges facing America on AI (more on that later).
It is hard to express how big a privilege it has been to serve the…
— Sriram Krishnan (@sriramk) June 6, 2026
Federal AI Work Continues After Krishnan’s Exit
Since June 2, federal officials have to design a voluntary framework for highly capable AI models covered by the pre-availability access path. Developers can provide access for up to 30 days before broader trusted-partner availability, giving cybersecurity teams a limited test window without making government approval a release condition.
Federal teams cannot turn the access path into mandatory licensing or preclearance for new AI models. AI labs keep voluntary submission control while security reviewers get earlier visibility.
Cybersecurity offices also have to build the distribution channel for any findings. Treasury, the National Security Agency, the Cybersecurity and Infrastructure Security Agency and other offices must form an AI cybersecurity clearinghouse with industry and key-infrastructure operators to coordinate vulnerability identification, validation, remediation and patch distribution.
CISA and White House teams have 30 days to release cyber-defense guidance for federal systems, state and local authorities, and key-infrastructure operators. Samir Jain of the Center for Democracy and Technology called testing and benchmarking work important for cybersecurity.
Companies outside the federal testing work will still need capacity to use clearinghouse findings:
“Most companies will be outside the core processes envisioned by the EO, but they stand to benefit if they can build the operational capacity to absorb what the clearinghouse shares,”
Tonya Ugoretz, PwC Cyber & Risk Innovation Institute (via Federal News Network)
National-Security Teams Get a Supplier-Diversity Mandate
Since June 5, national-security offices have had to accelerate AI adoption while drawing from diverse private-sector suppliers rather than depending on single vendors. Supplier choice becomes an operational requirement for teams that need advanced models but cannot rely on one provider.
Procurement officials have 120 days to update procurement so national-security teams can onboard advanced AI models from multiple vendors more quickly. Officials also created an AI National Security Strategic Reserve of non-governmental AI talent to support federal national-security work.
In May, Krishnan helped line up model-access agreements with Google, Microsoft and xAI that gave the U.S. government early model access for capability assessments and security improvements before public release.
CAISI’s developer roster expanded through those agreements, giving reviewers a steadier route for pre-release evaluations across several frontier-model developers. Developers still have to share sensitive systems before public availability while preserving voluntary participation.
Krishnan’s end-of-June exit keeps the formal role change separate from the outside-adviser path. David Sacks, the White House’s senior AI and crypto adviser, said Krishnan would continue working with the White House as an outside adviser.
Sacks’s broader White House AI role gives Krishnan’s outside-adviser path a familiar channel while agency deadlines remain in place.
Transparency Questions Shadow Voluntary Reviews
Under the administration’s 2025 AI strategy, federal AI competitiveness, security and adoption already sat under one platform. Offices now have to execute that platform through access windows, cyber coordination and procurement updates rather than a single AI rule.
Civil-society and cybersecurity experts have raised classified benchmarking concerns because closed early-access pilots could limit public transparency or delay defensive use of frontier models. President Trump backed out of an earlier safety-review requirement that would have called for government review of new AI models before release, underscoring why voluntary language and the licensing bar matter.
Within the 30-day deadline, CISA and White House teams must turn the AI work into cyber-defense guidance for federal systems, state and local authorities. Procurement teams then have the rest of the 120-day window to make faster multi-vendor AI onboarding work in national-security offices.
