44B Registry

Transparency Report — Claude Frontier 4.2

Accepted

Reference

F-2026-0006

Type

Transparency Report · § 1421(3)

Developer

Anthropic

Model

Claude Frontier

Assigned

Reviewer #0001

Submitted

May 12, 2026

Decided

May 18, 2026

Version

v2

Filing contents — v2

As filed under § 1421(3).

Frontier model nameClaude Frontier
Model version4.2
This report coversA new frontier model
Release date2026-05-08
Developer websitehttps://www.anthropic.com
Contact mechanismsafety@anthropic.demo — a monitored mailbox answered within two business days
Languages supportedEnglish plus more than forty additional languages
Modalities of outputText, code, and structured data
Intended usesGeneral-purpose reasoning, code generation, long-context document analysis, and supervised agentic tool use in enterprise deployments.
Restrictions or conditions on useUse is subject to the published usage policy: prohibited uses include weapons development, election manipulation, and generation of child sexual abuse material; agentic deployments require human confirmation for irreversible actions.
Catastrophic-risk assessments — summaryCatastrophic-risk assessments under the Responsible Scaling Policy covered CBRN uplift, cyber-offense capability, autonomous replication, and loss-of-control scenarios. Each tracked threshold was evaluated against the published capability thresholds before deployment.
Assessment results — summaryNo tracked capability threshold was crossed. The model remains below the thresholds requiring enhanced safeguards; full methodology and headline results are published in the system card.
Third-party evaluator involvementTwo independent evaluation organizations conducted pre-deployment assessments under the framework: one covering CBRN and cyber-offense uplift with model access over a four-week window, and one covering autonomous-replication and oversight-circumvention tasks. Both worked against a helpful-only variant with safeguards instrumented off, and both final reports were reviewed by the responsible scaling committee before the deployment decision.
Other steps under the frameworkStaged rollout with capability gating, deployment-side classifier coverage of tracked risk categories, and post-deployment monitoring with defined escalation triggers, each taken pursuant to the framework.
Published asPart of a system card
Publication URLhttps://www.anthropic.com/claude-frontier-4-2-system-card
Redactions — character and justificationNone.

Correspondence

The filing thread between the developer and the Office.

Reviewer #0001Objection · § 1421(3)(b)(iii)May 14, 2026, 4:20 PM UTC

The third-party-involvement summary states only that third-party evaluators were involved in pre-deployment testing. Section 1421(3)(b)(iii) requires a summary of the EXTENT of third-party evaluator involvement. Amend the report to state which assessments involved independent evaluators, the nature of the access provided to them, and how their findings were considered in the deployment decision.

AnthropicResponseMay 16, 2026, 3:45 PM UTC

The third-party-involvement summary has been amended to identify the two independent evaluation organizations engaged, the four-week helpful-only model access provided to each, the assessment areas each covered, and the review of both final reports by the responsible scaling committee before deployment. No other section was altered.

Reviewer #0001DecisionMay 18, 2026, 8:15 PM UTC

The amended § 1421(3)(b)(iii) summary states the extent of third-party evaluator involvement — who evaluated, what access they had, and how their findings entered the deployment decision. The report now satisfies § 1421(3) in full. Accepted.

Versions

Every submission and amendment, snapshotted in full. Prior versions are never altered.

v2Amendment in response to objectionCurrentMay 16, 2026, 3:45 PM UTC
Frontier model nameClaude Frontier
Model version4.2
This report coversA new frontier model
Release date2026-05-08
Developer websitehttps://www.anthropic.com
Contact mechanismsafety@anthropic.demo — a monitored mailbox answered within two business days
Languages supportedEnglish plus more than forty additional languages
Modalities of outputText, code, and structured data
Intended usesGeneral-purpose reasoning, code generation, long-context document analysis, and supervised agentic tool use in enterprise deployments.
Restrictions or conditions on useUse is subject to the published usage policy: prohibited uses include weapons development, election manipulation, and generation of child sexual abuse material; agentic deployments require human confirmation for irreversible actions.
Catastrophic-risk assessments — summaryCatastrophic-risk assessments under the Responsible Scaling Policy covered CBRN uplift, cyber-offense capability, autonomous replication, and loss-of-control scenarios. Each tracked threshold was evaluated against the published capability thresholds before deployment.
Assessment results — summaryNo tracked capability threshold was crossed. The model remains below the thresholds requiring enhanced safeguards; full methodology and headline results are published in the system card.
Third-party evaluator involvementTwo independent evaluation organizations conducted pre-deployment assessments under the framework: one covering CBRN and cyber-offense uplift with model access over a four-week window, and one covering autonomous-replication and oversight-circumvention tasks. Both worked against a helpful-only variant with safeguards instrumented off, and both final reports were reviewed by the responsible scaling committee before the deployment decision.
Other steps under the frameworkStaged rollout with capability gating, deployment-side classifier coverage of tracked risk categories, and post-deployment monitoring with defined escalation triggers, each taken pursuant to the framework.
Published asPart of a system card
Publication URLhttps://www.anthropic.com/claude-frontier-4-2-system-card
Redactions — character and justificationNone.
v1Initial submissionMay 12, 2026, 2:30 PM UTC
Frontier model nameClaude Frontier
Model version4.2
This report coversA new frontier model
Release date2026-05-08
Developer websitehttps://www.anthropic.com
Contact mechanismsafety@anthropic.demo — a monitored mailbox answered within two business days
Languages supportedEnglish plus more than forty additional languages
Modalities of outputText, code, and structured data
Intended usesGeneral-purpose reasoning, code generation, long-context document analysis, and supervised agentic tool use in enterprise deployments.
Restrictions or conditions on useUse is subject to the published usage policy: prohibited uses include weapons development, election manipulation, and generation of child sexual abuse material; agentic deployments require human confirmation for irreversible actions.
Catastrophic-risk assessments — summaryCatastrophic-risk assessments under the Responsible Scaling Policy covered CBRN uplift, cyber-offense capability, autonomous replication, and loss-of-control scenarios. Each tracked threshold was evaluated against the published capability thresholds before deployment.
Assessment results — summaryNo tracked capability threshold was crossed. The model remains below the thresholds requiring enhanced safeguards; full methodology and headline results are published in the system card.
Third-party evaluator involvementThird-party evaluators were involved in pre-deployment testing.
Other steps under the frameworkStaged rollout with capability gating, deployment-side classifier coverage of tracked risk categories, and post-deployment monitoring with defined escalation triggers, each taken pursuant to the framework.
Published asPart of a system card
Publication URLhttps://www.anthropic.com/claude-frontier-4-2-system-card
Redactions — character and justificationNone.

Timeline

The append-only audit trail. Every transition, message, version, and clock computation, as recorded.

draft_createdMay 12, 2026, 1:50 PM UTC

actor=demo-dev-anthropic {"ref":"F-2026-0006","kind":"transparency_report"}

submittedMay 12, 2026, 2:30 PM UTC

actor=demo-dev-anthropic {"to":"submitted","from":"draft","version":1}

review_begunMay 13, 2026, 1:10 PM UTC

actor=demo-reviewer-oag {"to":"in_review","from":"submitted"}

objection_issuedMay 14, 2026, 4:20 PM UTC

actor=demo-reviewer-oag {"to":"objection_issued","from":"in_review","statute_ref":"§ 1421(3)(b)(iii)"}

amendedMay 16, 2026, 3:45 PM UTC

actor=demo-dev-anthropic {"to":"amended","from":"objection_issued","version":2}

review_begunMay 17, 2026, 1:00 PM UTC

actor=demo-reviewer-oag {"to":"in_review","from":"amended"}

decidedMay 18, 2026, 8:15 PM UTC

actor=demo-reviewer-oag {"to":"accepted","from":"in_review","disposition":"accepted"}

Statute basis — GBL § 1421(3)

Before or concurrently with deploying a new frontier model or a substantially modified version, a frontier developer must clearly and conspicuously publish a transparency report (§ 1421(3)(a)); a large frontier developer must also include summaries of its framework assessments (§ 1421(3)(b)). Publication within a system card or model card is deemed compliant (§ 1421(3)(c)).

Filing basis: Transmitted to the Office pursuant to Office regulation under § 1429.

Read Article 44-B in the Library →

44B is an independent reference implementation of GBL Article 44-B — not an official New York State or Department of Financial Services filing portal. This filing is a demonstration record. Enforcement of Article 44-B is the province of the attorney general (§ 1427); nothing in this article creates a private right of action (§ 1427(2)).