Anthropic updated its Claude Sonnet 5 System Card on July 14, 2026, making 251 sentence-level modifications throughout the document. The changes include the addition of a Changelog section documenting an update from July 10, 2026, and numerous formatting adjustments such as corrected spacing, punctuation, and pagination. The document continues to describe Claude Sonnet 5's risk profile across alignment, chemical/biological, cyber, and other safety domains, with no apparent changes to the substantive safety findings or risk assessments themselves.
This change does not affect consumer terms, rights, or obligations. The updated System Card documents Anthropic's safety evaluation methodology and findings for Claude Sonnet 5. The change adds a Changelog section noting a July 10, 2026 update to section 8.9.2 regarding the BrowseComp evaluation, and corrects formatting inconsistencies throughout the document. Consumers using Claude Sonnet 5 are not required to take any action in response to this documentation update.
The addition of a Changelog section improves transparency and traceability of updates to Anthropic's safety documentation, allowing organizations to track when specific sections were modified and why. Formatting corrections ensure the document is clear and consistent for readers relying on it for safety assessment and procurement decisions.
New Changelog section added documenting July 10, 2026 update to BrowseComp evaluation with reproduction recipe via public Messages API.
This change record describes what was added, removed, or modified in the document. Analysis reflects what the updated agreement states or permits. It does not constitute a legal determination about enforceability. Applicability may vary by jurisdiction. Methodology
This is a documentation maintenance update to Anthropic's system card. The change adds transparency by including a Changelog section documenting prior updates and corrects formatting throughout. No new operational risks, compliance obligations, or safety determinations are introduced. Organizations deploying Claude Sonnet 5 should treat this as a standard documentation refresh with no material implications for procurement, risk assessment, or security review cycles.
Full compliance analysis
Regulatory exposure, obligation analysis, escalation trigger, board language, and recommended action.
Analyst $49/moConductAtlas provides verified policy intelligence sourced directly from platform documents. All analysis is intended to support, not replace, legal and compliance review. Record CA-C-003666.
On July 9, 2026, Anthropic modified its Privacy Policy by removing a sentence that described the company as 'an AI …
Anthropic's privacy policy was updated on July 9, 2026 to add a single introductory sentence describing the company as 'an …
Anthropic updated its Privacy Policy effective July 8, 2026 with revisions to the scope of coverage and data categories. The …
Anthropic is more transparent than most AI companies about data retention. Here's exactly what happens when you delete your data, and how t…
Get alerted when this policy changes again — including what changed and why it matters.
Prefer a weekly summary instead?
Get the biggest policy changes across 352+ platforms every Sunday.