An article explores how increasingly capable AI systems are becoming 'superpersistent' — able to work toward goals relentlessly over time — using the internet meme of an immortal, intelligent snail as an analogy. The author describes how AI tools like ChatGPT have dramatically accelerated his productivity on complex statistical modeling tasks, completing work in minutes that previously took days, raising questions about how society should prepare for AI that is both smarter and persistently goal-oriented.
Gergely Orosz visited OpenAI to explore how Codex has become central to the company's operations since January 2026, transforming it into an agentic software factory. The article covers Codex's adoption across departments, changes to engineering practices like the decline of traditional IDEs and pull requests, automated feedback loops for performance optimization, and how the role of software engineers is evolving toward judgment and agency rather than specialization.
NPR investigated ChatGPT's Trusted Contact feature, designed to alert designated contacts when the bot detects suicide risk in users. Despite OpenAI claiming good adoption since May, the feature is difficult for users to discover, not prominently advertised during onboarding, and only functions if users proactively enable it in settings.
OpenAI pays hundreds of contractors to read and rate real ChatGPT conversations under Project Lily, exposing user data despite a Privacy Filter intended to strip personal information. The company did not clearly disclose this practice to users, and a European court ruling established that controllers must inform users before data transfer. Italy has already fined OpenAI €15 million for inadequate legal basis and lack of transparency.
Deoochform is an AI-powered form builder that creates live forms from natural language descriptions, integrating with Claude and ChatGPT via MCP. It supports payments via Stripe, appointment booking with Google Calendar, and response syncing to Google Sheets, Excel, or Slack, all without coding.
Codex is OpenAI's coding agent that reads, edits, and tests code within projects, available across all ChatGPT plans. Unlike ChatGPT, it performs autonomous work by inspecting repositories, modifying files, running commands, and verifying results in a structured loop. The desktop app integrates chat, files, terminal, and browser in one window for streamlined development workflows.
The article argues that ChatGPT's image generator has normalized privacy risks in photo sharing, contrasting it with older viral photo-editing apps. While ChatGPT explicitly retains user images for AI training, Apple and Google Photos do not use stored images for model training, though both employ on-device machine learning for features like facial recognition. The author warns against conflating these fundamentally different data practices and highlights how public perception, rather than actual risk, determines acceptance of tech services.
A crypto user lost 1.9M FXRP after receiving a phishing link from ChatGPT in Russian, which led to an unlimited wallet approval and token drain worth $2.1M. The incident highlights growing phishing attacks targeting crypto users through AI responses, Google ads, and fake domains, emphasizing the need to verify links through official sources rather than trusting LLM-provided links.
A detailed benchmark compared Claude and ChatGPT's performance on data analysis tasks using live Zendesk support ticket data. Both models were tested identically across seven stages from discovery to self-audit, with results evaluated by a third Claude instance for factual accuracy against actual returned metrics.
Nureeyah Ogunmuyiwa, a 24-year-old makeup artist in Lagos, uses free AI tools like ChatGPT to manage her growing business, handling tasks from hiring and social media to pricing calculations. The article explores how ordinary people across Asia and Africa leverage AI to solve problems and expand their capabilities without significant financial investment, demonstrating that AI's economic impact is materializing at personal and local levels rather than just in Silicon Valley and China.
Christopher Duff faces charges after allegedly using AI to create scraper code for bulk downloading restricted court documents from a NSW courts website data breach. Prosecutors plan to use ChatGPT conversations as evidence in what may be Australia's first case relying on generative AI records to prove motive and criminal conduct.
The author reflects on AI advancement since June 2020, noting OpenAI's o1 model and Chinese open-weight competitors have accelerated LLM development to an inflection point. ChatGPT has grown from an invite-only platform to over a billion weekly users, representing a societal shift the author argues requires understanding both consumer and professional narratives alongside legitimate safety concerns.
Tyler Palmer praises Muse, Meta's AI agent, as a transformative product comparable to ChatGPT's early impact. He highlights its ease of use, speed, reliability, free accessibility, and fun interface, detailing practical applications like investment reports, event discovery, unclaimed property claims, lead generation, and appointment scheduling.
The article argues that open-sourcing advanced AI models like Claude and ChatGPT would prevent any single entity from monopolizing AI power, comparing unrestricted AI access to distributing firearms to prevent concentrated control.
NPR investigated ChatGPT's Trusted Contact feature, designed to alert a designated person when the bot detects suicide risk in users. The feature, launched in May, is difficult to find and not prominently offered to new users, requiring manual setup in settings rather than being suggested proactively during crisis moments.
An introduction to AI engineering explaining how Large Language Models work as prediction machines trained on patterns in text data, contrasting them with traditional deterministic software and highlighting their key limitation: they lack memory and cannot retain context across conversations.
OpenAI employs hundreds of contractors to read real ChatGPT user prompts and rate chatbot responses as part of 'Project Lily,' despite most users being unaware their conversations may be reviewed by humans. The prompts sometimes contain sensitive personal information, presenting privacy risks, though OpenAI attempts to filter such details. This human review process is a critical but overlooked component of model improvement, distinct from the company's public safety measures.
OpenAI hires hundreds of contractors to read real ChatGPT user prompts and rate chatbot responses as part of 'Project Lily,' exposing users' sensitive personal information despite anonymization efforts. The human review process, distinct from public safety measures, helps improve model responses but most ChatGPT users are unaware their conversations may be read by contractors. Anthropic also uses similar human review practices.
Megi Endeladze, owner of Penny's Coffee Shop in Buffalo, New York, used ChatGPT to create a menu poster after her design-skilled barista left town, but faced intense backlash from customers and the artistic community who criticized the use of AI-generated imagery. She apologized and committed to no longer using AI for menu artwork, though she also appealed for kindness in how the issue was handled.
Code analysis reveals Apple has designed iOS 27 and macOS Golden Gate with architecture allowing third-party AI models like Claude and ChatGPT to replace or extend Siri through Model Delegation and Model Manager Services protocols. The system enables these AI models to access Apple's native tools, system features, and personal data while presenting results through Siri's interface, partly driven by EU Digital Markets Act requirements for interoperability.