home.social

Alex Kleber a.k.a Privacy1st

  1. 🎭 “Which AI model are you?”
    “Whichever name the app told me to wear.”

    🚨 Our investigation into Newrock International’s “AI ChatBot - AI Pro Assistant” goes beyond suspicious marketing: we tested the Mac application and recovered its request and response data from the running app’s memory.

    The interface showed GPT‑6.2 Sol as the selected model.

    But the captured request specified gpt-6-luna. The matching completed response also reported gpt-6-luna.

    Most revealing was the instruction included in the request:

    “Your model identity is GPT-6.2 Sol.”

    When we asked the app for its exact model, it replied that it was GPT‑6.2 Sol. The captured instruction explains why that answer cannot be treated as independent confirmation of the model.

    ⚠️ Users should not have to inspect application memory to discover that the model label and the outgoing request disagree.

    The listing adds another misleading association through its subtitle: “Powered by ChatGPT+GPT-6 API”. ChatGPT is OpenAI’s chatbot product. Ordinary access to a GPT model through an API does not make another developer’s app ChatGPT. We did not observe ChatGPT sign-in or authorization to use a ChatGPT plan.

    A documented App Store review says the buyer subscribed believing the application was official. The developer’s own release history also records a previous content change at OpenAI’s request.

    🇻🇬 Apple’s German and Irish trader disclosures identify NEWROCK INTERNATIONAL LIMITED in British Virgin Islands. The same Apple developer account publishes the Mac app and an iPhone companion.

    Our controlled test covered the free Mac mode. It establishes misleading model presentation; it does not independently verify the intermediary server’s underlying model or premium delivery.

    The report includes screenshots, request/response extracts, the account diagram and an assessment against Apple’s review rules.

    📄 Investigation:
    appscamdb.com/reports/ai-pro-a

    💻 Mac app:
    apps.apple.com/us/app/ai-chatb

    📱 iPhone companion — not runtime-tested:
    apps.apple.com/us/app/ai-chatb

    👤 Developer account:
    apps.apple.com/us/developer/ne

    #AppStore #ChatGPT #AI #macOS #ConsumerProtection #AppScamDB

  2. 📸 Photos policy: no full Google Drive access. The shipped code packed a full-Drive permission. That is a surprisingly large photo album. 🚚

    New App Scam DB investigation: six Mac applications from MAN LI / Shield Apps, within one Apple developer account listing 15 products.

    🔎 The strongest finding concerns a privacy promise contradicted by compiled code.

    Quick Access for Google Photos says it does not request full Google Drive access. Its supplied executable includes a Google Sign-In branch that passes the full Drive permission.

    If that route is used and approved, the permission reaches Drive files beyond uploading photos. A separate, narrower Photos route also exists; we did not observe which branch a real user reaches or a live permission grant.

    🔐 We also traced Google authorization requests loaded into app-controlled browsers in Gmail, Docs and Photos. Google’s OAuth policy explicitly prohibits that architecture. The Gmail and Docs privacy policies nevertheless promise authorization following Google’s security best practices.

    Other documented findings:

    • Gmail’s App Store Privacy Policy button points to a non-resolving domain, although a different policy URL works.

    • Dropbox’s promotional artwork claims “#1 Dropbox App” without identifying a market, date or ranking source.

    • A Drive buyer reports paying about $20 and requests a refund. Positive reviews are also preserved; these are attributed complaints, not transactions we reproduced.

    • The four Google-named clients combine service names and familiar branding in a presentation we assess as misleading about endorsement. The authenticator separately uses Authy and Duo wording.

    These applications contain real API integrations alongside embedded web pages. We traced actual request construction and submission. That engineering does not make them official Google products, and it does not resolve the documented privacy and authorization discrepancies.

    The report includes original icon comparisons, a 15-app account diagram, promotional screenshots, reviews, exact code locations and the relevant Apple review rules. No password theft or fraudulent charge was established.

    Apple and Google should examine the production sign-in paths and disclosures.

    Full investigation:
    appscamdb.com/reports/shieldap

    Apple developer account:
    apps.apple.com/us/developer/%E

    Apps examined:

    Google Drive: apps.apple.com/us/app/id675481
    Google Docs: apps.apple.com/us/app/id675650
    Gmail: apps.apple.com/us/app/id675791
    Google Photos: apps.apple.com/us/app/id676848
    Dropbox: apps.apple.com/us/app/id675437
    Authenticator App – Authy, 2FA: apps.apple.com/us/app/id644431

    #MacAppStore #AppStore #Privacy #Google #CyberSecurity

  3. ⚠️ A suspected two-account scam comeback, OpenAI/Chrome branding, and a breadcrumb baked into the app. 🕵️🍞

    App Store plot twist: the compiled program came with a business card for another app.

    Our reporter alleges the same scammer used Muhammad Omie’s active account and Sameen Zaka’s reportedly suspended account. Here is the evidence we can actually show:

    🔎 Muhammad Omie publishes Ask Chat and Link Opener for Chrome & Edge. Both supplied applications share the same verified Apple signing team.

    🔗 Inside Link Opener’s main program, we found “Language learning app”, the marker “settings_langtalk” and the exact Langtalk App Store URL. The destination occurs in both Intel and Apple Silicon builds. The report shows exact byte positions and Apple Silicon instruction references.

    📚 Indexed Apple content names Sameen Zaka for Langtalk. A recovered September 2026 Apple page separately identifies Zaka’s developer account and shows its App for Teams Meeting advertising, including Microsoft Teams branding.

    ⚠️ This establishes a product-reference trail. Common ownership and an Apple suspension decision remain unverified. Earlier checks found the known Zaka product pages unavailable; the latest page requests were access-limited.

    The investigation also compares the current apps with official OpenAI/Chrome branding and examines privacy disclosures, subscription prices and relevant Apple review rules. Familiar branding deserves scrutiny when users could mistake an independent app for an official product.

    Full report, original images, connection diagram and executable evidence:
    appscamdb.com/reports/omie-two

    Current applications:

    Ask Chat:
    apps.apple.com/us/app/ask-chat

    Link Opener for Chrome & Edge:
    apps.apple.com/us/app/link-ope

    Apple developer accounts:

    Muhammad Omie:
    apps.apple.com/us/developer/mu

    Sameen Zaka — historical publisher; public profile persists:
    apps.apple.com/us/developer/sa

    Historical products examined:

    Langtalk:
    apps.apple.com/us/app/langtalk

    App for Teams Meeting:
    apps.apple.com/us/app/app-for-

    #Apple #AppStore #macOS #OpenAI #GoogleChrome #MicrosoftTeams #ConsumerProtection #scam #scammer #scamalert

  4. 🛡️ A.K Threat Hub — Daily Threat Report
    📅 19 November 2026

    On 19 November 2026, alleged data breaches at Binance, Banco Carrefour (Brazil), Aeroméxico (Mexico), Taxsee (Russian Federation), Central Bank of Nigeria (Nigeria) and Comerica Bank (United States) are claimed to have occurred on #DarkForums and #BreachForums. The largest claimed record count is 335.9M across 47 incidents. 192 incidents from 5 sources were logged, with 46 naming victim organizations.

    📊 Key numbers
    • 192 incidents, 44 critical or high
    • 25 countries, 13 sectors
    • 11 CVEs
    • Largest claimed leak: 335.9M records
    • Per day vs 18 November 2026: +5%

    🔥 Top incidents
    • #Binance — Data Breach, claimed by Darkleads2026, 335.9M records
    • #Banco #Carrefour (Brazil) — Data Breach, claimed by pingolitos, 11M records
    • #Aeroméxico (Mexico) — Data Breach, claimed by Eternal, 22M records
    • #Taxsee (Russian Federation) — Data Breach, claimed by loptrgod, 183M records
    • Central Bank of Nigeria (Nigeria) — Data Breach, claimed by FengZi_666, 30M records

    🎯 Most active actors: N0n (22), Cloudredhat (6), mk2_cloud (5), mk2clode (5)
    🌍 Most targeted countries: United States (23), Mexico (7), Germany (4), United Kingdom (4)
    🏭 Most targeted sectors: Financial Services (10), Transportation & Logistics (5), Government & Public Sector (4)

    📄 Full report — PDF, raw data, indicators CSV and MISP JSON:
    github.com/alexkleber/A-K-thre

    #ThreatIntel #CyberSecurity #DataBreach #Ransomware

  5. 🤖 First a ChatGPT-style knot. Now a ChatOn-style loop. An identity crisis with a recurring subscription.

    🚨 A highly downloaded US Mac app deserves more scrutiny—not more borrowed credibility.

    My new investigation covers Hira Amin’s “AI Chatbot・Ask AI Anything”, No. 13 in Apple’s US Top Free Mac chart when checked on 18 November.

    Here is what the evidence shows:

    🎭 Changing borrowed identities. The app’s archived icon closely imitates ChatGPT; its current artwork closely imitates ChatOn. Original images are compared side by side, alongside reviews describing ChatGPT purchase confusion.

    🔗 A documented development connection to another Apple account. Signed files from Hira Amin’s and Hadiqa Bashir’s apps name the same developer, reproduce a matching custom warning and share internal code labels. This is stronger evidence than a similar name or icon. It establishes a development link, not the legal ownership of both accounts.

    Hadiqa’s product is absent from the checked US, German and Irish Apple lookups, and the checked developer profile returns 404. Those results do not identify Apple’s enforcement decision or the reason for disappearance.

    🔐 Another company sits in the prompt’s path. The app’s privacy policy says submitted content passes through the developer’s Firebase server before reaching OpenAI. The policy says the server does not retain it. Users should still know that this independent operator processes their requests; this is not a direct conversation with OpenAI’s official ChatGPT app.

    💸 The listing offers $5.99 weekly and $49.99 yearly. Keeping the weekly subscription for a year has an estimated cost of $311.48.

    Apple should review the misleading presentation, purchase-confusion complaints and connected development—not treat download-chart visibility as reassurance.

    The report includes original branding comparisons, signed-file evidence, an account connection diagram, the data path and the relevant Apple review rules.

    Read the investigation:
    appscamdb.com/reports/chatbot-

    App:
    apps.apple.com/us/app/ai-chatb

    Historical Hadiqa app (absent from checked Apple lookups):
    apps.apple.com/us/app/id675295

    Apple developer accounts:
    Hira Amin: apps.apple.com/us/developer/hi
    Hadiqa Bashir: apps.apple.com/us/developer/ha

    #Apple #AppStore #MacOS #AI #ChatGPT #ConsumerProtection #Privacy #scam #scammer #chaton #openai

  6. 🗓️ Your calendar can remind you about meetings. Apparently it also needs to remind you to check who made the app. 💸

    Google Calendar Chrome made in Pakistan
    Cal AI clone made in Pakistan
    Apple - "null" brand protection

    My latest investigation examines two Mac apps using familiar brands to market different publishers’ products:

    🔵 Calendar App for Google, by Hafiz Muhammad Suleman, combines Google’s name with a close imitation of the blue “31” Calendar icon. The listing also switches between Calendar App for Google, Calendio and Calendy.

    🍎 AI Cal - AI Calorie Tracker, by Raeny Nadeem, reverses the established Cal AI name and echoes its white apple, dark background and scanning corners.

    🔗 These Apple developer accounts are connected through specific evidence: the supplied calendar build contains the exact privacy and terms URLs used by Raeny Nadeem’s Mac apps. The terms name Rainy Nadeem Apps.

    ⚠️ The calendar offers $7.99 weekly beside $34.99 yearly. Its build includes online account/event/note routes alongside a local-only privacy promise. The report documents the code; it does not claim observed data transmission.

    🇵🇰 The calendar is classified under Pakistan as an inferred development-region association, with the evidence explained in the report.

    Original icon comparisons, screenshots, code excerpts, an account diagram and the complete PDF:

    appscamdb.com/reports/calendar

    Apple developer accounts:

    apps.apple.com/us/developer/ha

    apps.apple.com/us/developer/ra

    Examined apps:

    apps.apple.com/us/app/calendar

    apps.apple.com/us/app/ai-cal-a

    #AppStore #macOS #GoogleCalendar #CalAI #Privacy #ConsumerProtection #scam #scams #Apple #Google

  7. 🛡️ A.K Threat Hub — Daily Threat Report
    📅 18 November 2026

    On 18 November 2026, alleged data breaches at Drive Medical involved 1,442,814 records in the United States. Alleged breaches at Hezbollah in Lebanon and Seaker in the Philippines also claimed large record counts. 183 incidents from 4 sources were reported.

    📊 Key numbers
    • 183 incidents, 20 critical or high
    • 14 countries, 10 sectors
    • 27 CVEs
    • Largest claimed leak: 830M records
    • Per day vs 17 November 2026: +3%

    🔥 Top incidents
    • #Drive #Medical - 1,442,814 records (United States) — Ransomware, claimed by Vortex, 1.4M records
    • #Hezbollah (Lebanon) — Data Breach, claimed by ROOT SYSTEM, 100K records
    • Hezbollah (Lebanon) — Data Breach, claimed by TonyXs911, 100K records
    • #Expedia (United States) — Data Breach, claimed by #NightBroker, 830M records
    • #Veritiv (United States) — Ransomware, claimed by Helix, 94.8M records

    🎯 Most active actors: #SurfCore (8), #WhiteMelly (8), #Matx1337 (7), #XH4X CYB3R (7)
    🌍 Most targeted countries: United States (6), Colombia (2), France (2), Guatemala (2)
    🏭 Most targeted sectors: Government & Public Sector (6), Media & Entertainment (4), Financial Services (3)

    📄 Full report — PDF, raw data, indicators CSV and MISP JSON:
    github.com/alexkleber/A-K-thre

    #ThreatIntel #CyberSecurity #DataBreach #CVE

Share on Mastodon

Enter the server where you have an account.