Top Risks

• THREAT FEED: extension ID independently confirmed malicious by TPCI chrome-mal-ids — AI Exporter - Save ChatGPT, Claude and Gemini chats to PDF, MD and more [credential-theft] + malext.io malicious_extension_sentry (+2.5, capped).
• Privacy policy fetched but scope_extension=false AND data_collection+third_party_sharing=true — admits broad data sharing without scoping to this extension (Privacy +10.0)
• brand_mention.is_impersonation=true (ChatGPT/Gemini) with gmail dev email and no verified business identity (Reputation +2.0)
• Broad host_permissions *://*/* on an AI content-scraping extension with cookies permission — can read/exfil any site session
• CSP sandbox allows unsafe-eval and unsafe-inline on script-src; DOM-XSS sink (innerHTML) found in bundled JS

Evidence

• privacy_policy_admits_third_party_sharing_without_scope api Policy fetched from saveai.net/privacy; scope_extension=false, data_collection=true, third_party_sharing=true → Privacy +10.0 (D rule)
• brand_impersonation_gmail_dev store brand_mention.is_impersonation=true (chatgpt,gemini); developer is mrsoniceman@gmail.com with no developer name; verified_publisher=true
• cookies_plus_all_urls manifest cookies permission + *://*/* host_permissions; ×1.2 amplifier applied — can read any site session cookie
• csp_unsafe_eval_unsafe_inline crx sandbox CSP contains unsafe-eval and unsafe-inline on script-src; dom_sink_innerhtml_userctrl in code_findings_raw → Code +2.0
• uninstall_url_hijack crx uninstall_url_hijack=true; monetization/tracking behavior on uninstall redirecting to 3rd party
• monetization_hits_google_analytics api threat_intel.monetization_hits=[google-analytics.com]; category=AI (not Adblock/Shopping) → Webstore +1.0 telemetry tier
• ai_extension_processing_page_content store AI/GenAI extension with content_scripts on 20 AI platform origins including chatgpt.com, gemini.google.com → Webstore +2.5
• free_webmail_no_developer_name store developer_email=mrsoniceman@gmail.com, developer_name=''; free webmail + no business name raises accountability risk
• threat_feed_membership api Extension ID listed by TPCI chrome-mal-ids — AI Exporter - Save ChatGPT, Claude and Gemini chats to PDF, MD and more [credential-theft] + malext.io malicious_extension_sentry; score 5.57 -> 8.07 (+2.5 cap 2.5).

Permissions Breakdown

• tabs medium Can enumerate open tabs and their URLs; moderate risk for content exfil.
• storage low Local data persistence; low standalone risk.
• cookies high Can read/write cookies across origins; paired with *://*/* enables session theft.
• contextMenus low UI affordance only; low risk.
• declarativeNetRequest medium Can modify/block network requests declaratively; moderate risk without webRequest.
• sidePanel low UI surface only; low standalone risk.
• *://*/* high Broad host access on all URLs; combined with cookies enables cross-site data access.

Pillar Scores

Bookkeeping