State & Local GovernmentIssue 312 records
State Government
Tonight's edition covers two states describing AI controls that rest on human review. Utah's governor signed an executive order on Oct 6 that makes AI literacy training mandatory for applicable executive-branch employees, requires human review for decisions affecting individual rights, and asks agencies to document how AI is used in public-facing processes, with the first progress report not due until July 2027. In Vermont, the digital services secretary told Compass Vermont that AI-suggested code fixes are applied only after people review them and that AI helped proofread ballots, but the agency isn't tracking how many issues were found or how severe they were. Neither source reports measured results.
- Evidence records
- 2
- Cross-source patterns
- 1
- Evidence classes
- 1 standards or public-body guidance1 independent reporting
- Outcomes
- 1 emerging1 mixed
- Source freshness
- 2 new this fortnight
- Research completed
- 2026-10-07
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Human review is the named control, but neither state is yet recording what the review catches
Utah's new order requires human-in-the-loop review for rights-affecting determinations and asks agencies to document AI's role, but sets no metrics and reports only to the governor starting July 2027. Vermont says every AI-suggested code change and every AI-checked ballot was reviewed by people, yet the agency doesn't track the number or severity of findings and gave no count of ballot errors caught. In both, human review is the safeguard on paper, and the record of what reviewers confirmed, rejected or missed doesn't exist yet. This is our interpretation from two sources.
Operating questionFor each AI use where a person reviews the output, where is the record of what the reviewer accepted, rejected and found that the AI missed?
Supporting evidenceOffice of Utah Governor Spencer J. Cox; Utah Department of Government Operations (DGO) and Division of Technology Services (DTS)Vermont Agency of Digital Services (Secretary Denise Reilly-Hughes), as reported by Compass Vermont
Evidence ledger
Full record · every source keeps its link and limitations
Utah executive order makes AI literacy training mandatory for state employees, requires human review for rights-affecting decisions, and asks agencies to document how AI is used
Gov. Cox signed Executive Order 2026-08 on Oct 6. It tells executive-branch agencies to avoid AI tools that don't follow the state's 'human-guided' and 'human-enhancing' principles. DGO must set safeguards against cyberattacks and 'unauthorized or rogue AI agents,' help agencies find uses that cut wait times 'while ensuring human-in-the-loop review for determinations affecting individual rights,' and use the State-Endorsed Digital Identity (SEDI) framework where applicable. Every agency must have applicable employees complete AI literacy training set by DTS and the HR division, look for cost-effective AI uses, and document how AI systems' inputs, outputs and roles in decisions fit into processes that affect the public. DTS must also train designated agency AI leads, prefer Utah-based training providers, and publish uniform AI implementation standards. The only reporting is an annual DGO progress report to the governor starting July 1, 2027. Higher education, the State Board of Education, the Attorney General, State Auditor, State Treasurer, the legislature and the courts are excluded.
Why it matters, evidence and limitations
- Why it matters
- Directly relevant to state CIOs and HR leaders: it turns a governor's AI initiative into binding duties for executive agencies, covering mandatory training, a documentation requirement for public-facing AI, and a human-review rule for rights decisions. It is also a template other governors may copy.
- Evidence and measured results
- Opened the signed PDF on the governor's site (curl, pdftotext) and read all seven sections. Cross-checked against the governor's Oct 6 news release and Utah News Dispatch and KSL coverage, which restate the order and add only official quotes.
- Limitations and uncertainty
- This is a policy instrument, not evidence of results. It contains no metrics, budget, deadlines for training completion, or definition of the human-review threshold. The governor's release carries an 'Oct. 30, 2026' dateline, which appears to be a typo; the order itself is dated Oct 6. Coverage so far restates official statements and includes no independent critique.
Vermont's digital services secretary says AI-suggested code fixes are applied only after human review and AI helped proofread ballots, but the state isn't tracking the number or severity of issues found
In written answers to Compass Vermont on Oct 5, Secretary Reilly-Hughes said developers have 'scanned dozens of applications' with Anthropic's tools under the state and local trial (June to December) and every application had improvement recommendations, 'many of which are minor.' The agency doesn't track the number or severity of issues centrally. AI suggestions are produced outside state systems and developers review and test changes before applying them. Vermont says it has never used Claude Security or Mythos; developers used Claude Code with Opus and Sonnet and also had unrestricted access to Fable, a Mythos-class model, and the answers don't say whether Fable was used on state code. Separately, the agency used Claude Opus to check ballot templates against public candidate information; staff reviewed every ballot, and CDAIO Raiche told StateScoop the job took three hours. No error counts were given. Both uses will be added to the state's public AI inventory in January 2027, and answers to lawmakers were due Oct 7.
Why it matters, evidence and limitations
- Why it matters
- This is the operator's first detailed written account after the Sep 30 Vermont oversight hearing, and it shows what a state can and can't currently say about an AI security trial: the controls (human review, no live access) are described, but outcomes aren't measured. The ballot use extends state AI to election administration support.
- Evidence and measured results
- Opened the Compass Vermont article (curl, BeautifulSoup), which quotes the secretary's written answers and links the Oct 2 StateScoop interview and Anthropic's Claude Code data policy. Vermont Daily Chronicle republished the same text. The Vermont legislature's JITOC page did not yet show the Oct 7 answers when checked.
- Limitations and uncertainty
- Operator-reported answers relayed by a single outlet; not the formal responses to the legislature, which were due Oct 7 and weren't posted when checked. No counts of vulnerabilities, fixes, or ballot errors caught; the three-hour ballot figure comes from a separate interview and has no baseline. Whether Fable touched state code is unanswered.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Standards or public-body guidance
- Normative or advisory guidance from a standards body or public institution.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.