Skip to main content

Security report for claude.ai

Anthropic PBC · Technology · United States

Medium risk. 6 critical or high findings in Security headers, GDPR outside view and 1 other check.

Security score

75/100

Risk levelMedium

0–100, higher is better. How it is calculated

Anthropic PBC operates the claude.ai domain, providing an AI assistant named Claude. The company is positioned as a technology innovator in AI conversational assistants, targeting developers, enterprises, and AI users. The website is professionally designed using modern frameworks such as React and Next.js, hosted via Cloudflare, and optimized for performance and mobile use. Security practices are strong, with HTTPS enforced and multiple security headers present, though DNSSEC is not enabled. Privacy and cookie policies are comprehensive and GDPR compliant, with consent mechanisms implemented. No direct contact information or security incident response details are publicly available on the login page. WHOIS data confirms the domain is legitimately registered to Anthropic PBC in the US since 2018, consistent with the company's founding and product timeline.

Scanned

NIS2 and GDPR posture

GDPR outside view
50/ 100 · Medium
Checks completed
7of 7

NIS2 Art. 21(2)

  1. (a) Risk analysis and security policies: Not visible from outside
  2. (b) Incident handling: One indicator
  3. (c) Business continuity, backups and crisis management: Not visible from outside
  4. (d) Supply chain security: Not visible from outside
  5. (e) Secure acquisition, development and maintenance, including vulnerability handling: Partly checked
  6. (f) Assessing the effectiveness of security measures: Not visible from outside
  7. (g) Cyber hygiene and security training: Not visible from outside
  8. (h) Cryptography and encryption: Partly checked
  9. (i) HR security, access control and asset management: One indicator
  10. (j) Multi-factor authentication and secured communications: Not visible from outside

This scan rated NIS2 with an earlier keyword method that is no longer shown; rescan for the per-article view. Risk management, continuity, supply chain, training, MFA, management oversight (Art. 20) and incident reporting (Art. 23) need internal evidence, and Guard cannot tell whether NIS2 applies to this organisation.

GDPR is indicative only, based on publicly visible signals: Articles 7, 12, 13 and 32 of Regulation (EU) 2016/679 and Article 5(3) of the ePrivacy Directive. Not a certification or legal assessment.

Priority findings

6 findings rated critical or high. Fix these first.

  1. Security headers

    Security headers are instructions your web server sends to every visitor’s browser. Missing ones leave visitors more exposed to attacks such as clickjacking and injected scripts.

    • High

      Missing Strict-Transport-Security header

      Forces HTTPS connections

      What to do: Add "Strict-Transport-Security: max-age=31536000; includeSubDomains; preload"

      • CWE-319
      • OWASP A04:2025
      • ASVS V3.4.1
    • High

      Missing Content-Security-Policy header

      Controls resources the browser is allowed to load

      What to do: Implement a Content-Security-Policy appropriate for your application

      • ASVS V3.4.3
      • OWASP Secure Headers: OWASP Secure Headers Project recommends Content-Security-Policy: default-src 'self'; form-action 'self'; base-uri 'self'; object-src 'none';…

  2. GDPR outside view

    These are the privacy signals visitors and data protection authorities look for under GDPR.

    • High

      No Privacy Policy found

      GDPR requires a clear and accessible privacy policy

      What to do: Add a comprehensive privacy policy page and link to it prominently

    • High

      No Cookie Consent Banner found

      GDPR requires explicit consent for non-essential cookies

      What to do: Implement a cookie consent banner that appears on first visit

    • High

      Third-party services without privacy policy

      Detected services: Google Analytics

      What to do: Document all third-party data sharing in your privacy policy

Need hands-on remediation? Email guard@offseq.com

Keep watching claude.ai

This scan found 16 issues across 6 checks. Monitoring re-runs those checks daily, emails you when something material changes and records each fix, so you can show progress.

Daily monitoring

Recommended

€25per month

  • TLS certificates and expiry
  • DNS records
  • SPF, DKIM and DMARC
  • Security headers
  • GDPR and NIS2 posture
  • Exposed services
  • Third-party trackers
  • Security score drift

Email alerts on material changes and an exportable JSON history of every change.

Billed monthly to your card. Cancel anytime. The full timeline opens with your unlock code.

Full report

€39.90one-time

  • The complete analysis as a PDF you can cite
  • Email delivery with an unlock code
  • Lifetime access to this report

Prices incl. 21% VAT. Secure checkout by Mollie; sold by SEQ SIA, Latvia.

Results by check

Each check with its score, findings and the evidence it collected.

How this score is calculated

The security score runs 0–100, higher is better. It is the plain average of the security checks that completed in this scan, each counted equally. A check that could not run is left out: it counts neither as a pass nor as a fail.

Counted in this score

SSL/TLS
75
Security headers
55
Email security
85
DNS health
85
Network security
100
GDPR outside view
50
Average
75

Not counted

  • NIS2 outside view: Re-weights evidence from the TLS, email, DNS, network and hygiene checks, so counting it would count that evidence twice. Shown on its own.
  • OffSeq Observations: An AI reading of the site, useful context but not reproducible evidence. Shown on its own.

Findings that cap the score at 49

Each of these is serious on its own, so while any is present the site cannot rate better than high risk, however good the other checks are.

  • The certificate is not trusted by browsers
  • The website is not served over HTTPS
  • A database, remote desktop, file-sharing or Telnet service is reachable from the internet
  • The page loads a script from a compromised source
  • The domain registration has expired
  • SPF authorises every server to send mail as the domain

Partial scans

When the site blocks Guard or most checks cannot run, the score covers only the checks that ran (usually DNS and email), is capped at 75 and is shown as partial, without a risk level.

Risk levels

  • Critical0–25
  • High26–49
  • Medium50–75
  • Low76–100

Technical appendix

Business context, site observations and technical notes gathered during the scan, and the standards behind each reference.

Business context

Business profile

Anthropic's business model centers on AI-as-a-Service, offering conversational AI tools and APIs for individual and team use. The company targets technology-savvy users and enterprises seeking AI assistance. The market position is that of an emerging leader focusing on safe and usable AI. The website content and branding are consistent and professional, supporting trustworthiness. The company uses reputable domain registration and hosting providers, indicating operational maturity. Marketing and analytics tools are used moderately, with Google Tag Manager and Sift detected. No suspicious or affiliate domains were found linked to the site.

Security posture

The website demonstrates a mature security posture with HTTPS, strong security headers, and no visible vulnerabilities or exposed sensitive data. The absence of DNSSEC is a minor gap. No explicit security policies or incident response contacts are published on the login page, which could be improved. The site does not appear to be behind a WAF or security challenge, allowing full content access. Overall, the security posture is strong with room for enhanced transparency in security policies and vulnerability disclosure.

Recommendations

  1. Enable DNSSEC to improve domain name system security and prevent spoofing.
  2. Publish a dedicated security policy and incident response contact information to enhance transparency and trust.
  3. Implement a vulnerability disclosure program or security.txt file to encourage responsible reporting of security issues.
  4. Consider adding verified company contact information on public pages to improve user trust and support.
  5. Maintain and regularly audit third-party scripts and analytics tools to minimize privacy risks and ensure compliance.
Site observations

Automated reading of the public site. Ratings are indicative.

Business

Company
Anthropic PBC
Description
Claude is an AI assistant developed by Anthropic, providing conversational AI services and tools for individuals and teams.
Key services
AI conversational assistant, API access to AI models, Team collaboration tools
Content quality
good
Branding
consistent

Technology

Stack
React, Next.js, Tailwind CSS, Cloudflare DNS
Frameworks
React, Next.js
Platforms
Web
Performance
fast
Mobile
good
Accessibility
basic
SEO
good

Security

Observed score
90/100
Good practices seen
  • HTTPS enforced
  • Client-side security headers present
  • No exposed sensitive data detected

Analytics and tracking

Services
Google Tag Manager, Sift
Tracking level
moderate
Privacy practices
good

Advertising

Marketing tools
Sift
Transparency
basic

Site quality

Design
good
User experience
good
Content relevance
good
Navigation
good
Professionalism
good
Trustworthiness
high

Key observations

  1. Website uses modern web technologies and frameworks
  2. Clear privacy and cookie policies with consent mechanisms
  3. No direct contact emails or phone numbers found on login page
  4. Strong security posture with HTTPS and security headers
  5. No evidence of suspicious or malicious content
  6. Content is focused on AI assistant services and related tools
  7. No WAF or blocking mechanisms detected
  8. WHOIS data is transparent and consistent with business identity
Technical notes

Notes

The website is built using modern JavaScript frameworks React and Next.js, styled with Tailwind CSS, and hosted on Cloudflare infrastructure. It uses multiple asynchronously loaded JavaScript chunks for modularity and performance. The site is mobile responsive and optimized for fast loading. SEO meta tags and Open Graph tags are properly configured for social sharing. Analytics and marketing scripts include Google Tag Manager and Sift for user behavior tracking and fraud prevention. No CMS is detected, indicating a custom or headless architecture. The technical implementation is robust and aligns with current best practices for web applications.
Standards and sources

Reference chips under each finding point to the standard, regulation or specification the finding relates to. They are taken from the official datasets below and mapped conservatively: a finding with no defensible link shows none. NIS2 and GDPR references mark where a finding is relevant evidence, not a compliance verdict. Implementing Regulation (EU) 2024/2690 binds digital infrastructure and digital service providers; for other entities it is the most detailed public reading of the Article 21 measures. WCAG, EN 301 549, European Accessibility Act and Web Accessibility Directive references name the requirement an automated accessibility rule relates to; they are not a conformance verdict.

Retrieved 2026-10-25

Earlier scans of this domain

  1. 75/100 +2Medium22 findingsView report
  2. 73/100Medium21 findingsView report
Reviews of this site

Share your experience with claude.ai. Every review is verified by email and published after moderation.

Overall rating
Select a rating
4000 characters remaining

We’ll email you to confirm your review and keep your details private.

Community rating

No published reviews yet. Be the first to review claude.ai.

Feedback on this report

Rate this report, add an optional comment, and tell us if you want a security specialist to follow up.

Overall report quality
Select a rating
2000 characters remaining

We only use your feedback to improve Guard reports. Contact details are never shared.