Live Signal · Updated 2026-11-16
Every Sovereign Beacon™ rating
Each rating below is an editorial opinion grounded in cited public reporting, dated and methodology-bound. Click any card for the full Sovereign Beacon™ Grade Card on sovereignware.org — the canonical archive containing the full claim, named sources, methodology version, right-of-reply contact, and corrections log.
Editorial disclosure: opinion · cited · dated · methodology-bound · right-of-reply offered
Aqara Camera Hub G3
The Aqara Camera Hub G3 offers privacy features like a physical shutter, local AI processing, and HomeKit Secure Video. However, multiple critical vulnerabilities severely compromise user privacy and security. In 2025, CVE-2025-65297 revealed the device automatically uploads sensitive telemetry and user data to remote servers without encryption or user consent, making it vulnerable to interception. Other 2025 vulnerabilities (CVE-2025-65295, CVE-2025-65293, CVE-2025-65291) expose it to malicious firmware installation and man-in-the-middle attacks. Furthermore, the official privacy policy (S0) explicitly states it "applies only to our online activities" and "is not applicable to any information collected offline or via channels other than this website," leaving device data handling opaque.
Opinion · Cited sources · As of 2026-11-16
SimpleLogin
SimpleLogin is a privacy-focused email alias service, acquired by Proton AG on April 8, 2022. Its privacy policy, updated in September 2025, states that all data is encrypted in transit via SSL/TLS and at rest, and it uses privacy-aware analytics (Plausible) that do not collect personal data. The service collects email metadata like sender, recipient, and IP, but explicitly states it "does not store your emails," processing content in memory and deleting it upon relay. Furthermore, SimpleLogin moved its legal domicile to Switzerland in January 2024, benefiting from strong Swiss privacy laws, and underwent a security audit by Securitum in March 2022.
Opinion · Cited sources · As of 2026-11-16
GrammarlyGO
GrammarlyGO, offered by Superhuman Platform Inc., presents privacy concerns due to its default data practices for many users. As of October 2026, Free, Premium, and single-user Pro accounts have user content used for AI training by default, requiring users to actively opt out. The privacy policy (S0, effective July 2026) confirms, "We also use information we collect to train our AI models," with an opt-out control. This practice, combined with past security vulnerabilities like the critical OAuth flaws in February-April 2023 that could have allowed account hijacking, and a March 2026 class-action lawsuit over the "Expert Review" AI feature, indicates a mixed privacy record. While Grammarly implements security measures like TLS 1.2 and AES-256 encryption and holds certifications, the default opt-out for AI training on user content for a large user base is a significant privacy risk.
Opinion · Cited sources · As of 2026-11-16
Maryland Online Data Privacy Act
The Maryland Online Data Privacy Act (MODPA), passed in 2024, significantly enhances consumer privacy protections by mandating strict data minimization, requiring controllers to "limit the collection of personal data to what is reasonably necessary and proportionate" for requested services. A key provision is the blanket prohibition on the sale of sensitive data, stating "A controller may not... sell sensitive data." Furthermore, consumers are granted the right to "opt out of the processing of personal data for purposes of... targeted advertising" and the sale of their data. The law also requires controllers to "establish, implement, and maintain reasonable administrative, technical, and physical data security practices" to protect personal data. Consumers also have the right to "delete personal data... unless retention... is required by law."
Opinion · Cited sources · As of 2026-11-16
Kagi Search
Kagi Search is a privacy-focused search engine operating on a subscription model, explicitly stating in its privacy policy (updated February 2025) that it "does not sell personal information or share it for behavioral advertising." The service employs strong technical safeguards, including HTTPS encryption and not logging searches with user accounts, a posture reinforced by a "Highly Secure" rating in an August 2022 security audit. In February 2025, Kagi further enhanced user anonymity with Privacy Pass authentication and a Tor onion service, making searches technically unlinkable. For AI services like Kagi Assistant, introduced in September 2024, conversation threads are deleted after one day by default, and third-party AI providers are selected to not save or use data for training. Server logs for debugging, such as Load Balancer Logs, are retained for a maximum of 7 days and are not linked to accounts.
Opinion · Cited sources · As of 2026-11-16
MyFitnessPal
MyFitnessPal is rated as a Threat due to its history of data insecurity and current privacy practices. In March 2018, the company disclosed a data breach affecting 150 million user accounts, exposing usernames, emails, and hashed passwords, some protected by weaker SHA-1. Furthermore, a class-action lawsuit filed in May 2025 alleges the company violated privacy rights by deploying tracking cookies for targeted advertising even after users opted out, with a federal judge allowing claims to proceed in February 2026. MyFitnessPal's privacy policy, effective March 18, 2025, states that while it does not "expressly 'sell' information," certain uses of functional and targeted advertising cookies "may constitute 'sales' or 'sharing' of personal information" under applicable laws. The policy details collection of sensitive health and fitness data, though it notes encryption in transit.
Opinion · Cited sources · As of 2026-11-16
AdGuard VPN
AdGuard VPN operates under a strict no-logging policy, explicitly stating it does not collect user activity, browsing history, connection timestamps, or IP addresses that could link an IP to a user, as detailed in its privacy policy (S0). The service employs AES-256 encryption and its proprietary TrustTunnel protocol. While it collects minimal data such as the number of bytes used and active VPN connections, this data is anonymized, not associated with locations, and retained for 90 days. In March 2025, AdGuard VPN for Android passed an independent security review by Leviathan Security Group. However, as of September 2026, the service had not undergone a comprehensive independent third-party audit of its no-logs policy or infrastructure, which is a notable concern for a VPN provider.
Opinion · Cited sources · As of 2026-11-12
Nitrokey 3C
Nitrokey 3C is a hardware security key designed to protect user data through open-source firmware and strong cryptographic mechanisms. The device itself does not store ordinary personal data but rather cryptographic keys and certificates, with FIDO credentials, OpenPGP, and PIV data protected by PIN encryption and the microcontroller's security mechanisms. Nitrokey GmbH, founded in 2015, emphasizes open-source hardware and software, allowing for independent security audits, such as the penetration tests by Cure53 in 2015. The company's privacy policy, updated in 2020 to comply with GDPR, explicitly states, "We don’t sell personal information to third parties." Although a PIV vulnerability (CVE-2025-25201) was identified in 2025, a firmware update (v1.8.1) was promptly released to address it, demonstrating commitment to security.
Opinion · Cited sources · As of 2026-11-12
Reddit poses a significant privacy threat due to its extensive data collection, sharing practices, and history of security incidents. In August 2018 and February 2023, Reddit experienced data breaches compromising user and internal data. The updated privacy policy, effective July 1, 2026, explicitly states that Reddit shares public content with "LLM providers who help us compile and summarize public content for relevant search results on Reddit," without a clear opt-out for users. Furthermore, the policy indicates, "We may share information if we believe disclosure is in accordance with, or required by, any applicable law, regulation, legal process, or governmental request," a practice highlighted by the May 2026 DOJ subpoenas for user information. These incidents and policies demonstrate a concerning approach to user privacy and data security.
Opinion · Cited sources · As of 2026-11-12
Mastodon
The official joinmastodon.org privacy policy, last updated in August 2026, states it collects only email addresses for newsletter sign-ups, allowing users to unsubscribe and delete their data. This policy, however, explicitly does not cover the privacy practices of individual Mastodon servers, which operate independently. A significant privacy concern for the Mastodon software is the lack of end-to-end encryption for direct messages, noted in November 2022, making them potentially visible to server administrators; end-to-end encryption is planned for 2027. The platform has also faced security challenges, including a critical vulnerability (CVE-2023-36460) discovered in July 2023 that could compromise servers, and a Mastodon server, Kolektiva.social, was compromised in July 2023 when an admin's electronics were seized. While the official site is transparent regarding its own data collection, the broader Mastodon ecosystem presents privacy risks due to its decentralized nature and current technical limitations.
Opinion · Cited sources · As of 2026-11-12
Hugging Face FHE Endpoints
Hugging Face FHE Endpoints are designed for privacy-first AI applications, leveraging Fully Homomorphic Encryption (FHE) to allow computation on encrypted data, ensuring user data remains private even during processing. The company explicitly states it "does not store customer data in terms of payloads or tokens passed to Inference Endpoints," with logs retained for only 30 days, and data in transit is secured with TLS/SSL encryption. While Hugging Face experienced a security incident in July 2026 involving a breach of internal infrastructure, the FHE technology inherently protects the confidentiality of data processed through these endpoints. The Privacy Policy, updated March 28, 2023, further states that the company "will not sell, rent or lease your Personal Information" and does not authorize third-party collection of PII for advertising without consent.
Opinion · Cited sources · As of 2026-11-12
Zoom
Zoom has a history of privacy and security concerns, leading to multiple legal actions. In November 2020, the FTC settled with Zoom over deceptive claims of "end-to-end, 256-bit encryption" for most meetings and installing software that bypassed browser security. While Zoom now offers E2EE, its privacy policy (S0) states that "Account owners" have significant control, allowing them to access and share "Recordings and Transcripts" and "In-Meeting/Webinar Messages" of participants, depending on their settings. Although Zoom clarified in 2023 that it "does not use any of your audio, video, chat... to train Zoom’s or its third-party artificial intelligence models" without consent, the policy also notes that Zoom shares personal data with "Vendors," "Marketing, Advertising, and Analytics Partners," and "Corporate Affiliates" for various purposes.
Opinion · Cited sources · As of 2026-11-12
Claude
Anthropic's Claude poses significant privacy risks due to its data collection and sharing practices. In August 2025, Anthropic shifted its consumer data policy to an opt-out model for using conversations to train AI models, extending retention for opted-in data to five years. The privacy policy states, "We may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings." Furthermore, in June 2026, the policy was updated to allow proactive sharing of conversation data with law enforcement based on a "good-faith belief" without a court order, as detailed in Section 3: "We may share personal data with government authorities, law enforcement... where... we have a good-faith belief that disclosure is reasonably necessary." This stance was underscored in March 2026 when a federal judge ruled that communications with Claude were not protected by attorney-client privilege due to its privacy terms. Additionally, in September 2026, a class action lawsuit was filed regarding the requirement for biometric scans for identity verification, which the policy confirms can include "facial geometry templates."
Opinion · Cited sources · As of 2026-11-12
Proton Pass
Proton Pass is rated yellow due to its strong end-to-end encryption for all stored data, including passwords and notes, ensuring Proton AG cannot access content. Its privacy policy, last modified May 2026, explicitly states, "We do not have the technical means to access the content of your encrypted emails, files, calendar events, passwords, or notes," and aims to "collect as little user information as possible." The service underwent a commendable security audit by Cure53 in July 2023 and another by Recurity Labs between January and April 2026, which rated its security "well above par." However, Proton AG, based in Switzerland, is legally obligated to comply with valid Swiss court orders. This was demonstrated in 2021 when a Swiss court compelled Proton to provide a user's IP address, and in 2026, when the FBI obtained customer billing information via Swiss authorities, highlighting that while content is secure, metadata can be compelled.
Opinion · Cited sources · As of 2026-11-12
Sync.com
Sync.com offers end-to-end encrypted cloud storage, asserting a zero-knowledge policy where it "does not access or process the files, content or documents you store" (S0). However, its Privacy Policy, updated September 15, 2026, clarifies that "some optional features may require specific files or selected content to be temporarily processed using a different encryption level" (S0), allowing Sync or its service providers temporary access to data for these features. This caveat, alongside concerns raised in November 2023 regarding shared links, was underscored when Sync.com acknowledged and addressed potential data leak vulnerabilities on links in October 2024. While Sync commits not to disclose personal information to third parties without consent, it collects account details, IP addresses, and logs interactions with files, sharing this with service providers or for legal processes (S0). Despite strong core encryption, these exceptions and recent security issues warrant a cautious rating.
Opinion · Cited sources · As of 2026-11-12
Experian
Experian, a multinational data broker, poses a significant privacy threat due to its history of data breaches and extensive data sharing practices. In October 2015, a breach exposed sensitive data, including Social Security numbers, of approximately 15 million T-Mobile users, leading to a 2022 multistate settlement. The company was fined $650,000 by the FTC in August 2023 for violating the CAN-SPAM Act by sending marketing emails without clear unsubscribe mechanisms. Its privacy policy (S0) states it collects 'Contact Information' and 'Device Information' and shares data with 'Group companies,' 'Suppliers,' 'Fraud prevention agencies,' and 'Public bodies, law enforcement and regulators.' A December 2024 policy update indicated increased sharing of subscriber transaction data for marketing purposes with affiliates and non-affiliated companies starting February 5, 2025, despite claims of using SSL and AES-256 encryption. Further concerns include a 2025 Dutch DPA fine for illegal processing and ongoing class-action lawsuits regarding selling 'trigger leads' and tracking web users.
Opinion · Cited sources · As of 2026-11-12
PimEyes
PimEyes is a facial recognition search engine that indexes 'fingerprints' of faces from publicly available internet sources, enabling users to search for individuals. This practice has drawn significant criticism and legal action, such as a complaint filed with the Hamburg Data Protection Authority in July 2020 and a lawsuit in Illinois in December 2023 alleging violations of the Biometric Information Privacy Act. While its privacy policy (updated April 23, 2026) states it 'does not engage in targeted searches for any particular individual,' it collects and processes biometric identifiers without explicit consent from data subjects. Furthermore, its opt-out mechanism requires users to provide a photo and an anonymized scan of their identification document, necessitating the submission of additional personal data to remove existing indexed information.
Opinion · Cited sources · As of 2026-11-12
Otter.ai
Otter.ai poses a privacy threat due to ongoing legal challenges and its data practices. In August 2025, class-action lawsuits were filed, alleging that Otter.ai recorded conversations without consent and used them to train its AI models, with a federal judge allowing core claims to proceed in August 2026. The privacy policy (S0) states it uses "Meeting and Uploaded Information" to "Improve and monitor the Services, including training our proprietary AI technology on de-identified audio recordings and on transcriptions," based on "Consent or Legitimate interests." For individual accounts, this AI training is enabled by default, requiring users to email to opt out, meaning user content is used for commercial AI model development without explicit opt-in consent. While Otter.ai completed a SOC 2 Type II audit in January 2023 and uses AES-256 encryption, the default data usage for AI training and the legal findings regarding third-party eavesdropping raise significant privacy concerns.
Opinion · Cited sources · As of 2026-11-12
Revolve.com
Revolve.com poses a significant privacy threat due to a major data breach and alleged tracking practices. In 2026, a data breach exposed customer payment information, including names, contact details, and payment card numbers, affecting at least 200,000 individuals, with Revolve failing to offer free identity theft protection. Furthermore, in July 2026, the company faced mass arbitration claims for allegedly using TikTok-related tracking technology to collect California users' data without consent. While Revolve's privacy policies (updated July 21, 2026) detail the collection of personal information, financial data, and internet activity, and note the use of cookies for targeted advertising, these practices, combined with the security lapse, demonstrate a failure to adequately protect user data.
Opinion · Cited sources · As of 2026-11-12
Vibe
Mistral AI encrypts all data at rest using AES256 and in transit via TLS 1.2+ as of August 2026, and states it does not "sell" or "share" consumer personal data for targeted advertising. However, for Vibe Free, Pro, and Team plans, user inputs and outputs are opted in for model training by default, though users can opt out. The privacy policy, updated September 2026, states user input and output data are retained until the account or conversation is deleted. A data breach occurred in May 2026, compromising a codebase management system, though Mistral AI stated hosted services and user data were unaffected. This default opt-in for training and the past breach warrant a yellow rating despite strong encryption and no data selling.
Opinion · Cited sources · As of 2026-11-12
Tuta Mail
Tuta Mail is an end-to-end encrypted email service committed to user privacy, operating under strict German data protection laws. Its client code is open source, allowing for public security audits. While a German court ordered monitoring of a single account for *future unencrypted* emails in November 2020, Tuta cannot decrypt end-to-end encrypted content. The service has promptly addressed security concerns, such as an XSS vulnerability in March 2021 and a client information leak (CVE-2024-23330) in November 2024. Tuta's privacy policy explicitly states it "protects all your data with strong encryption" and does not "plan to invest time and money into developing an AI that abuses your data," reinforcing its strong privacy-by-design approach.
Opinion · Cited sources · As of 2026-11-12
Washington My Health My Data Act
The Washington My Health My Data Act, signed into law on April 27, 2023, significantly enhances consumer privacy by requiring opt-in consent for the collection and sharing of consumer health data, as detailed in RCW 19.373.030. It mandates a separate, signed authorization for selling such data (RCW 19.373.070). The Act also introduced a ban on geofencing around healthcare facilities, effective July 23, 2023, to prevent tracking of sensitive health-seeking behaviors. Violations are enforceable by the Washington Attorney General and through a private right of action, providing robust protection for individuals' health information.
Opinion · Cited sources · As of 2026-11-12
EU Data Act
The EU Data Act, which entered into force on January 11, 2024, is designed to empower users by granting them greater control over data generated by their connected devices. It mandates data holders to make product data and related service data accessible to users and, upon request, to designated third parties in a structured, machine-readable format, often free of charge to the user. Article 12 of the Act explicitly addresses "Unfair terms in data access and use contracts," aiming to prevent unilaterally imposed terms that limit a data user's ability to choose a third party or use data for their own purposes. Furthermore, Article 10 outlines measures to preserve confidentiality, such as "aggregated personal data, anonymisation, pseudonymisation, and the provision of data access in a secure processing environment," particularly where trade secrets are involved. The Act also complements the GDPR, ensuring personal data protection is not diminished.
Opinion · Cited sources · As of 2026-11-12
Tinder
Tinder has a history of significant privacy and security issues. In 2026, Match Group, including Tinder, experienced a breach by ShinyHunters, exposing user identifiers and IP addresses, and the Irish DPC intended to fine Match Group for GDPR breaches related to data retention and transparency. The FTC also took action in 2026 for alleged sharing of user photos and location data without consent. The privacy policy, updated April 2026, states it collects "Face Data" (face geometry data) for verification and shares data with "Advertising Partners" for targeted ads, retaining data for a "safety retention window of three months following account closure." Earlier, in 2013, vulnerabilities exposed users' precise locations, and a 2017 study highlighted user unawareness of data disclosure, noting the policy stated personal information might not always remain secure.
Opinion · Cited sources · As of 2026-11-12
UK Online Safety Bill
The UK Online Safety Act 2023, enacted on October 26, 2023, poses a significant threat to user privacy, primarily through its potential impact on end-to-end encryption. The Act grants Ofcom powers to require platforms to scan for illegal content, a technical mechanism that critics, including major tech companies, warn could necessitate weakening encryption, as detailed in the Act's provisions under 'PART 7 OFCOM's powers and duties in relation to regulated services, CHAPTER 4 Information' (S0, S1). Furthermore, the Act mandates age verification or estimation for certain content, raising concerns about the collection of sensitive personal data, as outlined in 'PART 4 Other duties of providers... CHAPTER 1 User identity verification' (S0, S1). Despite government assurances in 2023 that these powers won't be used until technically feasible without compromising encryption, the provisions remain, creating a persistent risk to online privacy and security.
Opinion · Cited sources · As of 2026-11-12
Nextcloud
Nextcloud is designed for data sovereignty, with its software ensuring that 'no user data, especially personal data, is transferred to us,' as stated in its privacy policy (S0). The company has actively promoted GDPR compliance, releasing a GDPR Compliance Kit in May 2018, and was recognized by the EDPS in 2021 for empowering data control. However, Nextcloud GmbH experienced a significant internal data leak in May 2026, exposing approximately 367,000 records, including employee and client data, due to a hosting misconfiguration. Additionally, a 2FA bypass vulnerability (CVE-2024-37313) was identified in June 2024, though patches were promptly released. While the software's design prioritizes user privacy, these operational security incidents warrant a yellow rating.
Opinion · Cited sources · As of 2026-11-12
Netflix
Netflix is rated as a Red Threat due to its extensive data collection practices and a history of privacy violations and legal challenges. Its privacy policy (S0) explicitly states it collects "Usage information" including viewing history and voice inputs, and "Advertising information" such as device identifiers and IP addresses, which are used to support "Behavioral Advertisements" and shared with "Advertising Companies." In 2021, Netflix was fined by the South Korean government for collecting personal information without proper consent, and in 2024, the Dutch DPA fined Netflix €4.75 million for failing to provide clear information on data processing. Furthermore, multiple lawsuits were filed in 2026, including by the Texas Attorney General, alleging the company spied on users and collected children's data without consent, directly contradicting user expectations of privacy.
Opinion · Cited sources · As of 2026-11-12
EU Digital Services Act
The EU Digital Services Act (DSA) is a landmark regulation that significantly enhances user privacy and safety online. It entered into force on November 16, 2022, building upon the strong foundation of the GDPR, effective May 25, 2018. The DSA mandates enhanced transparency in online advertising, requiring disclosure of algorithms and personal data used for targeted ads, and explicitly prohibits targeted advertising based on sensitive data or to minors. Furthermore, it focuses on user control and data minimization, encouraging platforms to collect only necessary data and requiring clear, informed consent for data processing. This comprehensive framework ensures greater accountability for digital services and robust protection of personal data.
Opinion · Cited sources · As of 2026-11-12
Connected Vehicles
Connected Vehicles pose a significant privacy threat due to extensive data collection and sharing practices. Vehicles collect sensitive information like location, speed, and braking patterns, which is often disseminated to third parties. For instance, in March 2024, a New York Times investigation revealed General Motors shared customer driving data via its Smart Driver feature with data brokers, leading to a $220 million settlement in October 2026 for sharing data without proper consent. The S0 policy for the 'Connected Vehicles' website itself states that 'third-party ad servers or ad networks' use cookies and web beacons to collect IP addresses and track user activity for advertising purposes, over which 'Connected Vehicles has no access to or control.' The FTC also highlighted concerns in May 2024 regarding unlawful collection, use, and disclosure of personal data in connected cars.
Opinion · Cited sources · As of 2026-11-12
GitHub Copilot
GitHub Copilot poses a significant privacy threat due to its default data collection practices for individual users. A class-action lawsuit filed in November 2022 alleged privacy violations, including under the California Consumer Privacy Act (CCPA), concerning Copilot's training on publicly available code and potential personal data misuse. While many claims were dismissed by June 2024, GitHub's updated terms in March 2026 explicitly allow the use of interaction data (prompts, suggestions, code snippets) from individual Copilot Free, Pro, Pro+, and Max plans to train and improve AI models, unless users actively opt out. The GitHub General Privacy Statement (S0) confirms this, stating, "We use Personal Data to develop and improve our products, services, and technologies, including artificial intelligence and machine learning technologies," and that affiliates like Microsoft may also use this data for improvement. This default opt-in for training on user code and interactions, despite stronger protections for Business and Enterprise tiers, represents a substantial privacy risk.
Opinion · Cited sources · As of 2026-11-12
Privacy Badger
Privacy Badger is a secure, open-source browser extension developed by the Electronic Frontier Foundation (EFF) to protect user privacy by blocking third-party trackers. Since October 2020, it shifted to blocking trackers based on a pre-trained list to prevent fingerprinting vulnerabilities, enhancing its protective capabilities. The associated privacy policy (S0) states, "EFF does not sell or rent member, donor or website visitor information under any circumstances, and we do not share member, donor or visitor information with government entities except as compelled by law." It also sends Global Privacy Control (GPC) signals to websites. The EFF's commitment to user privacy, reinforced by its policy update in May 2026, ensures Privacy Badger remains a robust tool for digital rights.
Opinion · Cited sources · As of 2026-11-12
ElevenLabs
ElevenLabs poses a significant privacy threat due to its extensive data collection, default data usage for AI training, and recent security vulnerabilities. The privacy policy, updated October 9, 2026, states that "Usage Information" includes "mouse movements, clicks, keystrokes, data or text entered," which is highly invasive. Furthermore, user content is used by default to train AI models unless users opt out, and data generated about a user's voice is retained for up to three years after the last interaction. In October 2026, multiple users reported account hacks, with unauthorized plan upgrades and credit usage, leading to ElevenLabs Scale accounts being sold on dark web marketplaces in November 2026, indicating severe security flaws. These incidents, combined with broad data sharing for advertising and the potential for misuse highlighted by Senator Hassan in April 2026, warrant a red rating.
Opinion · Cited sources · As of 2026-11-12
KeePassXC
KeePassXC is an open-source, local-only password manager designed to protect user privacy by keeping all sensitive data encrypted on the user's device. Its privacy policy explicitly states, "At no time will any of your stored personal data be sent to us or a third party." The application employs strong cryptographic mechanisms like AES-256 encryption and Argon2d for master key derivation, ensuring data security. While the KeePassXC-Browser extension includes an optional automatic update check that contacts GitHub, the core application itself collects no telemetry. In 2025, KeePassXC received a First-level Security Certification (CSPN) from the French National Cybersecurity Agency (ANSSI), underscoring its robust security and privacy design.
Opinion · Cited sources · As of 2026-11-12
Grindr
Grindr is a severe privacy threat due to its extensive history of sharing highly sensitive user data and current policy clauses. In 2018, it was revealed Grindr shared users' HIV status and location, and a January 2020 Norwegian Consumer Council report found it shared detailed personal data with thousands of ad partners, leading to a 65 million NOK fine in December 2021, upheld in July 2024. The policy (S0, July 2026) explicitly states, "Your profile is public, so you should not include information that you want to keep private," and warns that displaying "Distance Information... could be used by some users to determine your exact location and to determine your identity." Furthermore, Grindr retains device identifiers and limited log information for up to two years after account deletion, as highlighted by an October 2023 EPIC complaint. Despite recent settlements like the September 2026 UK lawsuit, the app's fundamental design and data sharing practices, including with "advertising and marketing partners," continue to pose significant risks to user privacy.
Opinion · Cited sources · As of 2026-11-12
Kids Online Safety Act
The Kids Online Safety Act (KOSA) aims to protect minors online, originating from a 2021 Facebook leak that highlighted inadequate protections. It mandates platforms provide minors with default strong privacy settings, options to disable addictive features, and opt-out of personalized algorithmic recommendations. The bill also prohibits targeted advertising to minors without explicit parental consent, requiring default account settings to disable such advertising. However, despite explicit language stating it does not require age verification, critics argue KOSA could inadvertently push platforms towards more robust age assurance methods, potentially leading to increased data collection for age determination. As of March 2026, a revised KOSA, incorporated into the Kids Internet and Digital Safety (KIDS) Act, progressed in the House.
Opinion · Cited sources · As of 2026-11-12
Proton Drive
Proton Drive, launched publicly in September 2022, is a secure cloud storage service that employs end-to-end encryption (E2EE) for all user data, including file contents, folder names, and thumbnail previews. The encryption occurs on the user's device using open-source OpenPGP and elliptic curve cryptography, ensuring Proton AG cannot access the content. The privacy policy, last modified in August 2026, explicitly states, "Our overriding policy is to collect as little user information... as possible" and "We do not have the technical means to access the content of your encrypted emails, files...". Proton AG operates under strict Swiss privacy laws and publishes transparency reports, with a SOC2 Type II attestation completed in July 2025, further demonstrating its commitment to security and privacy.
Opinion · Cited sources · As of 2026-11-12
Vivaldi Browser
Vivaldi Browser is rated green due to its strong commitment to user privacy, highlighted by its "zero-tracking policy" and explicit statement in its privacy policy (last updated December 15, 2023) that it "do[es] not share or sell information to any third party." The browser implements end-to-end encryption for its Sync service, meaning Vivaldi cannot decrypt user data such as passwords or browsing history. While a unique user ID and anonymized IP address (last octet removed) are sent daily for aggregate active user statistics, browsing activity remains local to the device. This privacy stance has been consistent, with Vivaldi's CEO criticizing Google's data practices as early as 2017, and the browser further enhancing protection with a built-in ad and tracker blocker by 2021.
Opinion · Cited sources · As of 2026-10-09
System76 Laptops
System76 demonstrates a strong commitment to user privacy, explicitly stating in its 2026 privacy policy update that it "does not sell your data." The company's Pop!_OS operating system, often pre-installed on its laptops, is designed with privacy in mind, featuring full-disk encryption by default since 2018 and implementing Network Time Security (NTS) for encrypted time synchronization by 2022. Furthermore, the policy states that "System76 does not collect or store any information from Pop!_OS user installations," with minimal OS and hardware data used only temporarily for updates and connectivity verification, and no telemetry or error reports collected. This commitment extends to hardware, as seen in 2017 when System76 announced its intention to disable the Intel Management Engine (ME) firmware due to security concerns. The company also publicly voiced concerns in 2026 regarding proposed age verification laws, arguing they could compromise user privacy.
Opinion · Cited sources · As of 2026-10-09
Ring
Ring is a significant privacy threat due to a history of severe security failures and internal misuse of customer data. In May 2023, the FTC filed a complaint, leading to a $5.8 million settlement in June 2023, for allowing employees unrestricted access to customer videos and failing to implement adequate security, enabling hackers to take control of accounts. This followed an acknowledgment in January 2020 of employees improperly accessing customer videos, with one employee viewing thousands of recordings in 2017. Furthermore, in December 2019, over 3,600 account credentials were leaked after hackers exploited vulnerabilities. While Ring's privacy notice (S0) states it collects "video or audio recordings, live video or audio streams" and claims to "maintain administrative, technical and physical safeguards," these incidents demonstrate a failure to protect this sensitive data. A class-action lawsuit in June 2026 also alleges the "Familiar Faces" feature collects biometric data without consent.
Opinion · Cited sources · As of 2026-10-09
Montana Senate Bill 375
Montana Senate Bill 375 (SB 375), passed in 2025, primarily regulates the sale of edible hemp products containing THC and does not contain specific provisions related to user privacy. While categorized as a privacy law, its technical mechanism focuses on product regulation rather than data collection, sharing, or retention. Unlike other Montana legislation such as the Montana Consumer Data Privacy Act (MCDPA) enacted in 2023, which establishes consumer rights regarding personal data, SB 375 does not address the collection, use, or protection of personal information. Therefore, it neither enhances nor diminishes user privacy, leading to a neutral impact in this category.
Opinion · Cited sources · As of 2026-10-09
Llama.cpp
Llama.cpp, despite its local-first design, poses a significant privacy threat due to numerous critical security vulnerabilities. For instance, in April 2024, a remote code execution (RCE) vulnerability (CVSS 9.6) was found in `llama-cpp-python` due to improper handling of chat templates, allowing arbitrary code execution. Furthermore, in April 2026, CVE-2026-34159 identified another critical RCE in the RPC backend, enabling unauthenticated attackers to read and write arbitrary process memory. The project's security policy explicitly warns users under "Data privacy" that "To protect sensitive data from potential leaks or unauthorized access, it is crucial to sandbox the model execution," and advises against using the RPC backend in untrusted environments, acknowledging the inherent risks. These vulnerabilities can lead to unauthorized data access and breaches if not properly mitigated by users.
Opinion · Cited sources · As of 2026-10-09
Startpage
Startpage is a privacy-focused search engine that explicitly states it does not collect personal data for searches. Since 2009, it has not recorded IP addresses, and its policy confirms, "We don't record your IP address" and "We don't record your search queries." Technical mechanisms include anonymizing search queries before sending them to providers and using HTTPS encryption with SSL/TLS for all searches, a feature launched in 2011. Although acquired by System1 in 2019, Startpage maintains that its founders retain control over privacy implementations, and independent audits confirm its practices, including not building advertising profiles or selling personal data. A supplemental policy for accounts collects necessary billing and contact information, but explicitly states, "Having a Startpage account does not change how we treat your searches."
Opinion · Cited sources · As of 2026-10-09
X Corp.
X Corp. (formerly Twitter) has a history of privacy failures and a privacy policy that permits extensive data collection and sharing. In May 2022, Twitter agreed to a $150 million penalty with the FTC for misrepresenting that phone numbers and email addresses, collected for security, would not be used for advertising, a clear breach of trust. The current privacy policy (S0, effective January 15, 2026) explicitly states, "Based on your consent, we may collect and use your biometric information for safety, security, and identification purposes," and "We may collect and use your personal information (such as your biographical information, employment history, educational history...)" for job recommendations and sharing with employers. Furthermore, the policy allows sharing with "Third-party collaborators" who "may use it for their own independent purposes... including, for example, to train their artificial intelligence models." This broad data collection, coupled with past regulatory actions like the December 2020 GDPR fine for a data breach, demonstrates a significant threat to user privacy.
Opinion · Cited sources · As of 2026-10-09
Delaware Personal Data Privacy Act
The Delaware Personal Data Privacy Act (DPDPA), signed into law on September 11, 2023, significantly enhances consumer privacy by granting residents rights to access, correct, delete, and opt-out of the sale of their personal data, targeted advertising, and certain profiling activities. A key mechanism is the requirement for explicit consent to process "sensitive data," which broadly includes categories like racial origin, religious beliefs, health conditions, and status as transgender or nonbinary, as defined in § 12D-102(30) of the official text. The law, effective January 1, 2025, also uniquely applies to non-profit organizations. Further strengthening protections, House Bill 380, signed on September 2, 2026, lowers applicability thresholds and expands safeguards for minors, including prohibiting processing sensitive data of children under 13 without parental consent. While lacking a private right of action, enforcement is exclusively handled by the Delaware Department of Justice.
Opinion · Cited sources · As of 2026-10-09
BetterHelp
BetterHelp has a history of sharing sensitive user data for advertising purposes. In March 2023, the Federal Trade Commission (FTC) settled with BetterHelp for $7.8 million over allegations that it shared consumers' health data, including email addresses, IP addresses, and health questionnaire responses, with platforms like Facebook and Snapchat between 2017 and 2020, despite promising privacy. While BetterHelp's current policy states, "To be clear, we don't share any data or information you share with your Therapist with any Third Party advertisers," and uses "industry-standard encryption technologies" (SSL) for data in transit, the past actions demonstrate a significant breach of trust. The policy also notes that if users opt-in to "Advertising cookies and web beacons," some "Visitor Data" (excluding therapy data) may be shared for advertising, which in California may be considered a "sale" of personal information. Furthermore, data like "Clinical Health Record" and communications are retained for 10 years after the last login if therapy was started and erasure was not requested.
Opinion · Cited sources · As of 2026-10-09
Microsoft Pluton
Microsoft Pluton, announced in November 2020, is a security processor integrated directly into the CPU, designed to protect credentials, identities, personal data, and encryption keys from physical attacks by eliminating the communication channel to a discrete TPM. This technical mechanism significantly enhances on-device security. Microsoft's privacy statement (S0) generally commits to "protecting the security of your personal data" and using "a variety of security technologies and procedures" including encryption. However, Microsoft's broader privacy record includes a June 2023 FTC fine for COPPA violations related to collecting children's data without consent, and a January 2024 disclosure of a corporate email system data breach. While Pluton offers robust hardware-level protection, these incidents highlight ongoing company-wide privacy and security challenges.
Opinion · Cited sources · As of 2026-10-09
Fathom Analytics
Fathom Analytics is a privacy-first web analytics solution, designed for GDPR and ePrivacy compliance by not using cookies or fingerprinting for tracking (2019). It avoids persistent user profiles by generating site-specific visitor signatures using a daily rotating salt, IP address (stripped), and user-agent, without storing raw IP addresses. The privacy policy explicitly states, "We do not rent, buy or sell personal data from or to others," and "We process minimal personal data on your behalf" (IP address and User-Agent). In May 2020, Fathom undertook a comprehensive GDPR project, formalizing policies and appointing a Privacy Officer, and ensures EU visitor data is processed on European servers with IP addresses stripped within the EU. Data retention is limited to legal obligations.
Opinion · Cited sources · As of 2026-10-09
Acxiom
Acxiom, a data and marketing company, poses a significant threat to user privacy due to its extensive data collection and sharing practices. In 2002-2003, the company experienced a major data theft where 1.6 billion customer records were stolen from its FTP servers, highlighting severe security vulnerabilities. Its business model has repeatedly drawn scrutiny, including a 2003 complaint by EPIC to the FTC for sharing personal information without consent, and 2018 GDPR complaints by Privacy International alleging unlawful processing of data for profiling. The privacy policy, updated June 1, 2026, confirms its role in marketing data products, offering opt-out mechanisms for marketing and limiting sensitive personal information use, but this does not negate its fundamental practice of collecting and processing vast amounts of personal data for commercial purposes. Furthermore, a 2024 Bavarian data protection authority ruling implicated Acxiom in the alleged secret trading of personal data, and a 2026 lawsuit by the Texas AG alleged Netflix shared consumer profiles with data brokers like Acxiom.
Opinion · Cited sources · As of 2026-10-09
RoomGPT
RoomGPT is rated yellow due to its data collection and sharing practices, which present potential privacy risks for users. The privacy policy (S0) states that personal information, including name, email, and phone number, along with usage data like IP addresses, may be collected upon signup. While RoomGPT does not sell user data, it "may share your data with third-party service providers" (S0), who are obligated to keep information confidential but whose specific data handling for AI training is not explicitly detailed. An August 2026 review highlighted potential privacy concerns related to uploading personal photos to the platform, and discussions in September 2026 emphasized the need for users to confirm whether inputs are used for model training (S4, S5). Although generated images are not saved to user accounts as of August 2026, the ambiguity regarding the use of sensitive input data for AI model training and sharing with third parties warrants a cautious approach.
Opinion · Cited sources · As of 2026-10-09
Firefox Focus
Firefox Focus is designed with privacy in mind, offering features like tracker blocking and clearing browsing data. Its privacy notice (S0, effective April 22, 2025) states that "Your browsing data is cleared from your device whenever you close your browser" and that "Focus and Klar process technical and settings data... on your device." However, Mozilla has faced scrutiny for its data practices. In 2024, Mozilla was criticized for sending data, including advertising ID and IP address, to Adjust GmbH from Firefox for Android and iOS to track ad installs without prominent disclosure, leading to the Adjust SDK's removal by August 2024. While the app offers strong privacy controls, the collection of "minimal technical and interaction data" for daily usage pings, retained for 25 months, and past third-party data sharing incidents prevent a fully green rating.
Opinion · Cited sources · As of 2026-10-09
DuckDuckGo AI Chat
DuckDuckGo AI Chat is designed with strong privacy protections, ensuring conversations are not used for AI model training and personal information like IP addresses is removed before being sent to underlying AI model providers. The service explicitly states it does not save or share chat history, with recent chats stored locally on the user's device by default. For longer-term storage, users can enable encrypted Sync & Backup, where decryption keys remain solely on user devices. Furthermore, DuckDuckGo introduced private chat sharing in October 2026, encrypting shared content on the user's device without DuckDuckGo accessing the decryption key, reinforcing its commitment to user-controlled data privacy.
Opinion · Cited sources · As of 2026-10-09
Shein
Shein is a significant privacy threat due to a history of data mismanagement and regulatory non-compliance. In June 2018, its then-parent company, Zoetop, suffered a data breach affecting 39 million accounts, leading to a $1.9 million settlement in October 2022 for failing to adequately secure customer data, despite its policy stating, "We maintain physical, electronic, and procedural safeguards designed to guard and prevent misuse of your personal information." Furthermore, the company's broad data sharing practices, outlined in its policy as "We may disclose and share your personal information with the parties as described below," have drawn scrutiny, including a February 2026 Texas AG lawsuit alleging potential accessibility of consumer data to the Chinese government. Shein also faced a €150 million fine from France's CNIL in September 2025 for violating cookie laws by placing advertising cookies without user consent, contradicting the policy's "Your Choices" section which implies user control. These repeated failures and ongoing investigations highlight a systemic disregard for user privacy and data protection.
Opinion · Cited sources · As of 2026-10-09
Telegram
Telegram's privacy posture has significantly weakened, particularly after its September 2024 privacy policy update. The company now explicitly states it "may disclose your IP address and phone number to the relevant authorities" if a user is suspected of criminal activities violating Terms of Service, a major shift from its previous stance. This change followed CEO Pavel Durov's arrest in August 2024 and was reflected in January 2025 transparency reports showing data disclosure for over 2,200 users in 2024. While "Secret Chats" offer end-to-end encryption, default "Cloud Chats" use client-server encryption, meaning Telegram retains access to message content, and a 2016 API flaw also exposed user data.
Opinion · Cited sources · As of 2026-10-04
Flock Safety Falcon & Condor Cameras
Flock Safety's surveillance systems force constant video, situational telemetry, and audio captures straight to a unified vendor cloud. The hardware functions with zero standalone or isolated network profiles, meaning all device data feeds cross-agency searching without a distinct warrant. Investigations by civil liberties groups confirm updated contractual stipulations extend expansive corporate data brokerage permissions and drop previous 'no-sale' protections, pushing the hardware strictly to Critical.
Opinion · Cited sources · As of 2026-06-24
Pine64 PinePhone Pro
The PinePhone Pro places ultimate hardware control back in user hands. Beneath its removable rear case sit six physical DIP switches capable of disconnecting circuit lines for the cellular modem, Wi-Fi/Bluetooth chips, microphone, front camera, rear camera, and headphone jack. Because the entire platform runs on community-audited Linux operating systems with zero vendor-mandated cloud services or baked-in diagnostic telemetry pipelines, it achieves absolute local data sovereignty.
Opinion · Cited sources · As of 2026-06-24
Threema Mobile Messaging App
Threema uses an architecture that generates a unique, anonymous 8-digit Threema ID for each client instance, ensuring messaging maps and profile networks are isolated from cellular identity trails. Group structures and contacts are handled entirely client-side rather than centralized on remote servers. While its business model relies strictly on up-front application sales (avoiding ad-tech monetization vectors), it requires centralized routing infrastructure to sync and transit payload streams, capping its final tier at Clear.
Opinion · Cited sources · As of 2026-06-24
Adobe Creative Cloud Applications
Adobe's modern software stack features mandatory integrations with its centralized Creative Cloud environments. Deep investigations into Adobe's service agreements reveal that standard accounts automatically opt creators into global 'Content Analysis' loops, allowing remote server arrays to scan active project binaries to refine generative algorithms. Because these diagnostic telemetry systems run out of the box and verification layers depend entirely on closed corporate trust, it drops to the Warning tier.
Opinion · Cited sources · As of 2026-06-24
AnythingLLM Framework
AnythingLLM functions completely isolated from vendor network arrays. All contextual documents, chat interaction histories, and embedded databases are maintained within the local disk space of the client machine. Anonymous telemetries are fully transparent and can be permanently deactivated within the initial settings window. Because it supports completely auditable open-source validation without mandatory external callbacks, it achieves a Sovereign rating.
Opinion · Cited sources · As of 2026-06-24
Midjourney Generative AI
Midjourney lacks any local computing architecture, channeling all client image synthesis loops through commercial cloud rendering clusters. Its basic operating policies permanently retain image inputs, vector generations, and personal interaction profiles for internal platform optimization and modeling metrics. Because the pipeline relies on un-verifiable black-box processes, features no safe defaults, and functions entirely through a mandatory cloud network, it ranks at Critical.
Opinion · Cited sources · As of 2026-06-24
SearXNG Metasearch
SearXNG establishes a proxy barrier between individual seekers and commercial web indexing structures. It discards user profile paths completely, omitting contextual search logs and script trackers. While private self-hosted instances can approach a sovereign footprint, general web deployment sites still require clear routing over standard web clouds to query data sources, capping the platform state firmly at Clear.
Opinion · Cited sources · As of 2026-06-24
LinkedIn Professional Network
LinkedIn utilizes a cloud framework optimized to chart behavioral networks. Policy reviews disclose that the platform deploys standard opt-out configurations for corporate AI modeling, automatically indexing user communications and resume assets to cultivate machine networks. Because the baseline system configurations operate unsafely without a manual adjustment, its placement settles at Warning.
Opinion · Cited sources · As of 2026-06-24
European Union Artificial Intelligence Act (EU AI Act)
The EU AI Act safeguards user autonomy by declaring automated biometric and predictive scoring architectures as 'unacceptable risks,' rendering them entirely prohibited across the European market. It establishes rigid machine-readability and transparency mandates on models to block covert content extraction. However, because it relies on centralized regulatory databases and allows specific law enforcement cloud exemptions in tracking public threats, it does not achieve local offline status, placing it at Clear.
Opinion · Cited sources · As of 2026-06-24
US FISA Section 702 (RISAA Reauthorizations)
Section 702 serves as a sweeping domestic spy tool, compelling commercial electronic communication service providers to pipeline data signals to intelligence databases without individualized warrants. Modern reauthorizations (RISAA) have actively expanded the types of businesses forced to cooperate with state collection nodes. With persistent, documented internal agency violations and zero end-user verification vectors, it maps unequivocally to Critical.
Opinion · Cited sources · As of 2026-06-24
YubiKey
YubiKey is a physical cryptographic security key that executes authentication operations entirely on-chip within its secure element. The hardware has zero network capabilities, does not contain tracking firmware, and does not require a cloud connection to function. Rigorous third-party testing for its FIPS 140-2 certification in 2023 verified that its cryptographic keys are physically un-extractable, ensuring absolute sovereign possession.
Opinion · Cited sources · As of 2026-06-07
Purism Librem 5
The Purism Librem 5 is a security-focused smartphone that operates completely on local user-owned hardware with no mandatory cloud integrations. It features physical hardware kill switches that disconnect the camera, microphone, Wi-Fi, Bluetooth, and cellular baseband from power. Independent reviews and Privacy Guides have confirmed that its baseband processor is isolated from the main CPU, ensuring no background data harvesting can bypass user controls.
Opinion · Cited sources · As of 2026-06-07
PrivateGPT
PrivateGPT is an open-source AI application designed to provide a completely offline conversational interface for document analysis. Since the application, embedding database, and LLMs execute entirely on local CPU/GPU hardware, no telemetry or prompt history is transmitted to any cloud servers. The full source is auditable on GitHub and the project explicitly documents that no remote API requests are made during ingestion or inference, ensuring physical and cryptographic custody of sensitive files.
Opinion · Cited sources · As of 2026-06-07
Mullvad VPN
Mullvad VPN is an application and routing service designed with absolute privacy-by-design principles, requiring no email address or personal details to register. Although it routes traffic through a cloud infrastructure, the service is built to prevent user data logging or profiling. In April 2023, a Swedish police search of Mullvad's offices demonstrated that the company stores zero customer data, as the authorities were unable to retrieve any user logs.
Opinion · Cited sources · As of 2026-06-07
Bitwarden
Bitwarden is an open-source credential manager that secures all user data with zero-knowledge, client-side encryption before syncing to the cloud. Its software architecture is fully open for public audit, and the company routinely publishes third-party security assessments. An independent code audit completed in 2024 confirmed that the cryptographic implementation is robust and that Bitwarden has no ability to decrypt or profile customer vaults.
Opinion · Cited sources · As of 2026-06-07
ProtonMail
ProtonMail provides zero-access encrypted email storage, meaning messages are encrypted client-side before being written to Proton's Swiss-based servers. An independent cryptographic audit completed in late 2024 confirmed that the web interface handles encryption processes correctly without leaking private keys. Because it relies on a trusted external operator, it does not achieve true sovereign status, but its structural design prevents data profiling.
Opinion · Cited sources · As of 2026-06-07
Brave Search
Brave Search is an independent search engine that operates on its own proprietary web index without building user behavioral profiles or tracking queries. The platform serves search results without utilizing third-party cookies or scripts that monitor cross-site behavior. In mid-2023, Brave officially removed all remaining Bing search API dependencies, making its search results completely independent of major surveillance-based search providers.
Opinion · Cited sources · As of 2026-06-07
macOS with iCloud
Apple's macOS operates locally but automatically prompts users to sync highly sensitive personal files, photos, and keychains to iCloud servers under default settings. In January 2023, the French regulator CNIL fined Apple 8 million euros for failing to obtain explicit user consent before presenting personalized App Store advertisements. Achieving a secure environment requires users to manually activate Advanced Data Protection and disable telemetry tracking.
Opinion · Cited sources · As of 2026-06-07
Perplexity AI
Perplexity AI is a conversational search engine that depends entirely on continuous cloud extraction and real-time indexing of web content. Its default terms allow the collection of user search habits, prompt history, and diagnostic data to train internal models. In mid-2024, a Wired investigation revealed that Perplexity bypassed standard robots.txt exclusions and scraped private sites without authorization. Since the platform's core model depends on aggressive data crawling and default logging, user configuration cannot guarantee complete privacy.
Opinion · Cited sources · As of 2026-06-07
Temu
Temu is an e-commerce application that operates on a business model driven by aggressive data extraction and consumer tracking. A June 2023 US Select Committee report highlighted major concerns regarding Temu's parent company, PDD Holdings, and its potential to harvest comprehensive mobile device data. In 2024, multiple consumer protection lawsuits were filed alleging that the application accesses contact lists, locations, and search histories without authorization.
Opinion · Cited sources · As of 2026-06-07
Apple Vision Pro
Apple Vision Pro introduces severe surveillance risks due to its vast array of always-on sensors, including multiple depth sensors, microphones, and 12 cameras. Although eye-tracking data is processed locally, the device continuously maps physical environments and tracks body movements in three dimensions to blend digital content with the physical world. This comprehensive environmental data collection creates unprecedented privacy concerns, as detailed 3D mapping could inadvertently expose intimate personal details, socio-economic status, and sensitive household items to third parties or malicious actors if exploited.
Opinion · Cited sources · As of 2026-06-05
Framework Laptop
Framework laptops are engineered with a heavy emphasis on user privacy, the right to repair, and open-source compatibility. By default, they feature physical hardware kill switches that completely sever power to the built-in microphone and webcam, ensuring they cannot be accessed surreptitiously by malware or background processes. Additionally, they are "Linux-first" friendly, offering pre-built options and official support for privacy-respecting operating systems like Ubuntu and Fedora, allowing users to entirely avoid the invasive telemetry associated with mainstream proprietary software.
Opinion · Cited sources · As of 2026-06-05
Amazon.com
Amazon's e-commerce platform relies on aggressive cross-site tracking and opaque data processing practices to build extensive behavioral profiles for targeted advertising. This systemic non-compliance with data protection standards culminated in a historic €746 million ($887 million) GDPR fine issued by Luxembourg's National Commission for Data Protection (CNPD) for failing to obtain explicit user consent before processing personal data. The site persistently monitors user activities across the web, capitalizing on vast amounts of personal metadata to consolidate its market advantage while ignoring fundamental privacy principles.
Opinion · Cited sources · As of 2026-06-05
DuckDuckGo
DuckDuckGo operates as a privacy-first search engine that actively mitigates cross-site tracking by refusing to store search history, collect IP addresses, or build personalized user profiles. Unlike mainstream competitors, it anonymizes every search request and delivers results without employing algorithms designed to harvest behavioral data. The platform further protects users by offering built-in tracker-blocking tools and HTTPS encryption enforcement, effectively neutralizing third-party tracking scripts that normally profile users as they browse the broader web.
Opinion · Cited sources · As of 2026-06-05
Ollama
Ollama is an open-source, local-first framework that allows organizations to run large language models (LLMs) entirely on their own hardware. It processes all prompts and data locally with zero external API calls or network requests after the initial download, ensuring complete data sovereignty and eliminating cloud leak risks. Its network-isolated operation prevents third-party tracking, making it a gold standard for privacy-compliant enterprise AI deployments.
Opinion · Cited sources · As of 2026-06-05
OpenAI ChatGPT (Consumer & Free Plans)
By default, OpenAI's consumer-facing ChatGPT and Plus accounts actively utilize user chats, uploaded files, and personal inputs to train future models. Unless a user manually navigates to settings to explicitly opt out, sensitive data is permanently processed on cloud servers and integrated into third-party datasets. Additionally, conversations are retained on OpenAI's servers for up to 30 days even with history disabled, representing a persistent risk of intellectual property exposure.
Opinion · Cited sources · As of 2026-06-05
California Privacy Rights Act (CPRA / CCPA)
The California Privacy Rights Act (CPRA) is widely considered the strongest consumer data privacy law in the United States. It provides robust, enforceable consumer rights, including the right to delete, correct, limit the use of sensitive personal information, and opt out of automated decision-making. Significantly, it is one of the very few state statutes to offer a limited private right of action for data breaches and features a dedicated enforcement watchdog, the California Privacy Protection Agency (CPPA).
Opinion · Cited sources · As of 2026-06-05
Utah Consumer Privacy Act (UCPA)
The Utah Consumer Privacy Act (UCPA) is heavily criticized by digital rights advocates for prioritizing business interests over consumer safety. It fails to mandate fundamental data minimization principles, does not provide consumers with a private right of action, and severely limits enforcement authority to the state Attorney General. Due to its high monetary applicability thresholds and extensive business exemptions, the law is rated as highly deficient and effectively acts as a shield for corporate data-harvesting practices.
Opinion · Cited sources · As of 2026-06-05
Virginia Consumer Data Protection Act (VCDPA)
Passed as one of the earliest state-level frameworks, the VCDPA is rated RED by privacy advocates because it was heavily influenced by industry lobbyists and serves as the template for weak state-level legislation. The law lacks a private right of action, giving sole enforcement power to the state Attorney General, and includes broad exemptions for financial institutions and insurance companies. This creates a regulatory environment where corporate compliance is easily navigated without substantial changes to tracking behaviors.
Opinion · Cited sources · As of 2026-06-05
Iowa Consumer Data Protection Act (ICDPA)
The ICDPA is rated RED as it is one of the weakest state privacy laws passed in the United States. Unlike stronger statutes, it denies consumers the fundamental rights to correct inaccuracies in their collected personal data or to restrict automated profiling. With zero private right of action, no rulemaking authority for a dedicated regulator, and an extraordinarily generous 90-day cure period for non-compliant businesses, the law acts more as a rubber stamp for industry data practices than a consumer protection mechanism.
Opinion · Cited sources · As of 2026-06-05
eufy Home Security Cameras
In January 2025, the New York Attorney General secured a $450,000 settlement after finding that eufy's video streams were not end-to-end encrypted and could be accessed without authentication. This lack of proper encryption and risk of foreign surveillance has led to investigations by U.S. and Australian authorities in early 2026.
Opinion · Cited sources · As of 2026-05-31
GrapheneOS on Google Pixel
GrapheneOS is a hardened, open-source Android operating system that provides a tracking-free environment by disabling background telemetry and advertising IDs by default. It implements full-disk encryption and hardened app sandboxing to isolate and protect user data at the silicon level.
Opinion · Cited sources · As of 2026-05-31
TikTok
Despite transitioning to U.S.-based management in January 2026, TikTok's updated privacy policy has introduced expanded collection of precise location data. Additionally, the app collects and profiles metadata from all user interactions with its AI-powered features, creating extensive privacy tracking risks.
Opinion · Cited sources · As of 2026-05-31
Signal Private Messenger
Signal remains the gold standard for secure messaging in 2026, offering audited, end-to-end encryption by default for all communications and utilizing Sealed Sender technology to hide metadata. It is operated by a non-profit foundation, does not back up data to third-party clouds, and maintains a strict zero-data-retention policy.
Opinion · Cited sources · As of 2026-05-31
Replika AI Companion
In May 2025, Italy's data protection authority (Garante) fined Replika's developer, Luka Inc., 5 million euros for serious GDPR violations, including processing sensitive personal data without a valid legal basis. The platform permanently logs highly intimate chats, fails to implement age verification, and lacks transparent data retention notices.
Opinion · Cited sources · As of 2026-05-31
Jan.ai
Jan.ai is an open-source, local-first personal AI chat application that runs entirely on the user's hardware by default. It processes all conversations and documents locally, ensuring that no sensitive data or chat history is transmitted to or stored on third-party cloud servers.
Opinion · Cited sources · As of 2026-05-31
Tractor Supply Company
Tractor Supply Company's website was fined a record $1.35 million by the California Privacy Protection Agency in September 2025 for serious CCPA violations. The site failed to honor consumer opt-out requests, ignored Global Privacy Control signals, and shared user tracking data with third-party advertising networks.
Opinion · Cited sources · As of 2026-05-31
Plausible Analytics
Plausible Analytics is an open-source, GDPR-compliant web analytics platform designed to offer detailed metrics without using cookies or collecting personally identifiable information. It minimizes data exposure and is hosted on secure European infrastructure, ensuring complete compliance with global privacy standards.
Opinion · Cited sources · As of 2026-05-31
