From the State Government edition of October 8, 2026
Vermont posts the signed Anthropic Evaluation Agreement: $100,000 credit for Claude Security and Claude Code, defensive cyber use only, six-month term
State of Vermont and Anthropic, PBC (agreement posted by ADS to JITOC) · State AI cybersecurity evaluation contracting · Vermont statewide (customer: State of Vermont)
- Publisher
- Vermont General Assembly / Agency of Digital Services
- Original publication
- 2026-10-07
- Source retrieved
- 2026-10-09
What happened
Alongside its Oct 7 answers, ADS posted the signed Anthropic Evaluation Agreement titled “Claude Security (Government).” Customer is the State of Vermont. Evaluation Services are “Claude for Enterprise including Claude Security and Claude Code with cyber defense runbooks,” with a Credit Amount of $100,000.00 USD for a six-month Evaluation Period from the Effective Date. Approved use is solely for defensive cybersecurity purposes in customer-owned or authorized systems (scanning, triaging, validating, remediating vulnerabilities); general software development is barred unless Anthropic permits in writing. Services are provided “as is”; Claude Security is labeled a Beta Service; liability is capped at $1,000 or fees paid. Government terms allow public-disclosure-law sharing of the agreement with notice to Anthropic. This contract text sits beside ADS’s same-day statement that Vermont is not using Claude Security.
Why it matters
Shows the actual instrument behind a free state SLTT AI security trial: credit amount, scoped defensive uses, beta disclaimer, and liability cap. Useful for peer states comparing vendor evaluation templates. Jurisdictional limit: Vermont’s Bulletin 3.5 treatment is separate from the vendor form.
Evidence and measured results
Opened the PDF posted on the JITOC Oct 7 documents page via curl and pdftotext. Compared service description to ADS’s written statement that Claude Security is not used.
Limitations and uncertainty
Effective Date and wet-ink signature dates are not clearly extractable from the text layer (form fields). Title and Evaluation Services include Claude Security even though ADS says it is not using that product—scope on paper may exceed stated practice. Liability and as-is terms are vendor-favorable; no measured security outcomes appear in the contract.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-10-08; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Oversight bodies want the signed AI evaluation agreement, not only hearing testimony.
- Stakeholders
- CIO/CISO
- Procurement / Secretary of Administration
- Legislative IT oversight
- Vendor account team
- Discovery questions
- Which products named in the evaluation form are actually authorized for use?
- What is the credit balance and evaluation end date?
- Are as-is/beta and liability terms acceptable to counsel?
- How will public-records disclosure of the agreement be handled?
- Value hypothesis
- Transparent evaluation agreements with scoped defensive uses help agencies trial AI cyber tools without overstating production readiness.
- Potential engagement
- Contract-and-practice reconciliation memo for one SLTT-style evaluation.
- Unsupported claims
- That Vermont production systems are covered by Anthropic warranties
- That Claude Security is in active Vermont use
- Guaranteed security outcomes from the credit
Pre-sales engineering
Role takeaway
- Fit
- Relevant when an evaluation form lists multiple AI cyber products and the agency will enable only a subset.
- Architecture and integration
- Map form services (Claude Security, Claude Code, runbooks) to actual IDE/plugin and hosting path; disable or block unused products.
- Prerequisites
- Authorized-use matrix
- Hosting choice (native vs Azure/Bedrock/Vertex)
- Credit metering
- Constraints
- Defensive-use-only credit
- Beta status for Claude Security
- Six-month term
- Security
- Enforce approved-use boundaries; do not point credits at general development.
- Proof of value
- Export of enabled products vs agreement list
- Credit-utilization report
- Sample defensive finding handled under approved use
Delivery
Role takeaway
- Work
- Publish agreement to oversight, operate within defensive scope, monitor credits, close or convert before term end.
- Dependencies
- Counsel acceptance of as-is/beta terms
- Inventory/procurement close-out
- Security ops capacity
- Ownership
- ADS contracting and security leads.
- Skills and adoption
- Developers and security staff need clarity on allowed vs disallowed uses under the credit.
- Governance checkpoints
- Signed agreement posted to oversight
- Authorized-product list approved
- Mid-term credit and scope review
- Acceptance criteria
- Proposed: authorized-product list signed by CISO/CDAO
- Proposed: monthly credit burn reported
- Proposed: unused contracted products explicitly disabled
- Risks
- Paper scope exceeds practice, creating oversight confusion
- Credit runs out mid-remediation
- Beta/as-is terms leave residual risk with the state
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Contract contemplates Claude Security and Claude Code with cyber defense runbooks in Anthropic-native or marketplace environments (Azure AI Foundry, Bedrock, Vertex). ADS separately says it uses Claude Code via VS Code and Azure-hosted LLM and does not use Claude Security.
Governance
Who approves, reviews and stays accountable for outcomes?
The form is an evaluation credit agreement, not a standard Vermont Bulletin 3.5 zero-dollar contract package (no AA-14/Attachment C in the posted file). Public-records carve-outs are built in. Beta and as-is terms shift risk to the state customer.
Security and privacy
What data, permissions and controls need testing?
Approved use is limited to defensive cyber on systems the customer owns or is authorized to test. The posted agreement does not itself state data-retention settings for individual models; those appear in ADS’s Q&A.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Limited direct relevance; the agreement restricts credits to cyber-readiness work rather than general development, which shapes which staff may use the tools under the credit.
Procurement
What should contracts, pricing and exit terms secure?
Zero-cash $100k credit evaluations still create binding terms (use limits, beta status, liability caps). Agencies should reconcile contract product lists with actual authorized tools before legislative disclosure.
Operating model
Which teams own the service once it runs?
Credit depletion ends access; either party may terminate for convenience. Operational planning should track token burn (ADS reports ~10% used) against the six-month window.
Publication history
- 2026-10-08State Government · Issue 332 resources
Stable resource ID: vermont-anthropic-evaluation-agreement-claude-security-20261007