OpenAI Contractors Manually Review Real User Chats in Project Lily
OpenAI has recruited hundreds of third-party contractors to manually review real user dialogues with ChatGPT. The work forms part of project Lily and gives reviewers direct access to prompts and responses that may include personal, medical, or other confidential details.
Reviewers are tasked with briefly summarizing each user query, comparing four model-generated answers, and scoring them on a scale from one to seven. They also flag issues such as excessive sycophancy, attempts by the model to appear human, and inappropriate use of emojis.
Although account names are masked and conversations pass through a Privacy Filter, OpenAI confirmed to 404 Media that the filter does not always remove all personal data. Reviewers may additionally see blocks of user memories containing prior conversation context and approximate location information.
Contractors are hired through external firms including Crossing Hurdles and Mercor, with some specialists earning more than $50 per hour. A similar human-evaluation process is used by Anthropic for conversations with Claude.
Under OpenAI policy, chats from personal plans may be used to improve models. Users can disable this by navigating to Settings → Data Controls and turning off the “Improve the model for everyone” option. The setting applies account-wide but does not retroactively remove previously processed data. Temporary chats are excluded from training, though they may still be reviewed for abuse detection.
Related articles
GrapheneOS Setup Guide: Configuring Pixel Phones for Corporate Surveillance-Free Daily Use
This comprehensive engineering guide explains how to deploy GrapheneOS on supported Google Pixel devices to eliminate corporate telemetry collection. It follows three core principles: rejecting proprietary ecosystems, applying Zero Trust through cryptography and open-source audits, and enforcing strict compartmentalization via isolated user profiles. The tutorial covers official installation via the Web Installer, basic owner profile hardening with PIN shuffling and automatic reboot, and the use of Obtainium for direct FOSS app management from GitHub repositories. Detailed recommendations include privacy-focused tools such as KeePassDX, Aegis Authenticator, AmneziaVPN, Signal, and Fossify applications, along with VPN kill-switch configuration. Regional profiles are created for sandboxed Google Play, Aurora Store, RuStore, and Huawei AppGallery to safely run banking, marketplace, and social apps without cross-profile tracking.
Following the White Rabbit: Developer Builds Custom Rust VPN PAYPHONE Using QUIC and Obfuscation to Evade Detection
A Russian developer has released PAYPHONE, an experimental IPv4 VPN written entirely in Rust that uses QUIC datagrams and optional TLS-over-TCP transport with custom obfuscation. The project aims to provide an alternative to AmneziaWG and Xray/VLESS+REALITY stacks that are commonly used to bypass Russian internet filtering. The article details the full packet path from TUN interface through a 16-byte PAYPHONE header, session management with Ed25519 tokens, and multiple post-launch bugs including MTU miscalculations, self-routing loops on macOS, and timer lifetime issues in Tokio. Key technical choices include RFC 9221 datagram support to avoid head-of-line blocking for multiplexed TCP flows and token-bucket rate limiting tied to subscription tokens. The author also describes route monitoring every 400 ms and interface-bound sockets to prevent the tunnel from swallowing its own control traffic.
WhatsApp Introduces Parental Controls for Teen Privacy Settings
WhatsApp, owned by Meta (recognized as an extremist organization and banned in Russia), has rolled out new parental control tools for family accounts. Parents can manage privacy settings, group participation, channel access, status visibility, and Meta AI usage for teens, but cannot read personal messages due to end-to-end encryption. All controls are voluntary and require joint setup with the teenager, protected by a single PIN code that prevents easy reversal of restrictions. Notifications alert parents when teens join or leave groups or when group sizes change significantly. Separate options cover channel usage, viewable statuses, and audience controls for teen posts. Meta AI access can be set to a standard 13+ mode or a stricter Limited Content mode with undisclosed restrictions. The company plans to expand these features gradually based on family feedback while maintaining encryption protections.
Can Wi-Fi Owners See Your Google Search History? HTTPS, DNS, SNI and ECH Explained
A viral social media video sparked widespread concern that Wi-Fi owners could view users' search history and visited sites simply by knowing the router password. Security experts from Cybernews and Surfshark clarified that modern HTTPS encryption prevents reading of actual search queries or page content. However, metadata such as DNS requests, SNI fields in TLS handshakes, and device MAC addresses remain visible to the network administrator. The introduction of Encrypted Client Hello (ECH) under RFC 9849 aims to hide domain names, yet Russian authorities have blocked many ECH-enabled connections since November 2024. Corporate or school-managed devices with installed root certificates represent the main real-world exception where full traffic inspection is possible. VPNs hide destinations from the local router but transfer visibility to the VPN provider. The article emphasizes that password-protected Wi-Fi grants access only to connection metadata, not browser history.