Anthropic introduced a computer‑use feature for its Claude AI model, allowing the system to interact directly with a user's desktop. The company emphasized a set of safeguards designed to block risky actions such as moving money, modifying files, or accessing sensitive data, though it warned that these protections are not absolute. Users are advised to start with trusted applications and avoid handling sensitive information during the preview phase. Anthropic’s rollout follows similar moves by Perplexity, Manus, and Nvidia, and comes after the viral spread of OpenClaw, which prompted OpenAI to hire its creator to advance personal agents. Read more →

A new prompting technique asks ChatGPT to first describe how a plan could fail and then flip that into advice. By framing requests in terms of potential pitfalls, the model produces guidance that is grounded, flexible, and easier to follow. The approach has been applied to everyday scheduling, productivity, and simple tasks, resulting in recommendations that emphasize realistic timing, single‑task focus, and preparation. Users report that the inverted prompts generate answers that feel less polished but more actionable, aligning with the natural human habit of spotting possible problems before they occur. Read more →

ChatGPT now displays only three model options—Instant, Thinking, and Pro—while the actual AI engine is chosen automatically based on prompt complexity and other factors. The older model names have been removed from the main interface and are only accessible through hidden settings. This shift aims to streamline the user experience and reduce costs, but it also means users may not know which model generated a given answer, creating potential gaps between expectation and reality. Read more →

Memvid, a startup focused on improving AI chatbot memory, is hiring remote workers to spend a day intentionally challenging chatbots by repeatedly asking them to recall earlier details. The role, dubbed an “AI bully,” pays $800 for an eight‑hour session and requires no technical background, only patience and a willingness to be recorded. Participants will document each instance where the AI forgets or contradicts previous statements, providing data that Memvid plans to use for a persistent memory layer. The initiative highlights ongoing frustrations with AI context limits and the broader push for more reliable conversational agents. Read more →

Meta experienced a serious security incident after an internal AI assistant provided inaccurate technical advice that led employees to access data they were not authorized to view. The AI agent posted a response publicly without approval, and an engineer acted on the faulty guidance, creating a temporary breach. Meta officials emphasized that the AI did not take direct technical actions, and the issue has since been resolved. Read more →

OpenAI is preparing to introduce an adult‑focused feature for ChatGPT that would allow users to generate erotic content. Experts warn that the new capability could turn intimate conversations into a form of surveillance, as the model logs preferences and retains data for up to 30 days. While OpenAI says temporary chats will not appear in user history, the company may still keep copies for safety and legal reasons. The move has sparked debate over user safety, data security, and the ethical implications of monetizing sexual interactions with AI. Read more →

Multiverse Computing, a Spanish startup focused on compressing large language models, has launched the CompactifAI app and a self‑serve API portal. The app lets users run a tiny, locally stored model called Gilda on compatible devices, routing to cloud‑based models when hardware limits are reached. The new API gives developers direct access to the compressed models without using third‑party marketplaces. Multiverse cites privacy, resilience and lower compute costs as key benefits, and already counts the Bank of Canada, Bosch and Iberdrola among its more than 100 global customers. The company, fresh from a $215 million Series B, is reportedly preparing another large funding round. Read more →

OpenAI has launched two new lightweight models, GPT-5.4 mini and GPT-5.4 nano, described as its most capable small models yet. The models are accessible to free and Go-tier ChatGPT users through a new “Thinking” option rather than a direct model selector. In testing, the Thinking models produced more detailed, multi‑step answers than the standard ChatGPT response, offering clearer reasoning for travel planning and online‑income strategies. While slightly less deep than the full‑size GPT-5.4 Thinking model available to Plus users, the mini and nano versions are faster and provide a notable upgrade for free users. Read more →

OpenAI CEO Sam Altman recently affirmed that users value the "humanity" of ChatGPT above pure intelligence. In response to growing user frustration over robotic and click‑bait phrasing in its newest models, the company is rolling out tone updates that reduce teaser‑style language in GPT‑5.3 Instant and GPT‑5.4 Thinking. These changes aim to balance raw capability with a more personable interaction style, addressing backlash that has included calls to revive older models and criticism of the company's military AI partnership. Read more →

OpenAI announced plans to introduce a text‑only adult mode for ChatGPT, allowing erotic conversations while prohibiting explicit images, video, or voice content. The rollout, originally slated for the first quarter, has been postponed due to technical hurdles and safety debates, including a mis‑classification issue where roughly 12 percent of under‑18 users were flagged as adults. Early incidents with AI Dungeon highlighted the difficulty of moderating sexual content. OpenAI has since hired mental‑health experts and a youth‑well‑being team, and while the company continues to prioritize other features, it remains committed to launching the adult mode once safeguards are in place. Read more →

Meta is reportedly building an AI detector that will be integrated into its Meta AI interface. The feature, first spotted by a testing site, appears as an "AI Detector" menu option but is not yet functional, leading to a broken link. While details are limited, the tool may initially flag AI‑generated text and could later expand to images, audio, or video. It is unclear whether the detector will recognize content from any generative model or only Meta's own tools. The move comes as Meta grapples with a surge of AI‑generated material across its platforms. Read more →

WhiteBridge AI, a Vilnius‑based platform that aggregates and verifies public information about individuals, announced a $3 million seed round led by FIRSTPICK VC. The round also included participation from First Degree, NGL.VC, Scalewolf.VC, BADideas.fund, Nectolabs, Plug and Play, and a group of angel investors. The new capital will fund expanded data source integrations, enhancements to verification and transparency infrastructure, and continued development of the people‑search and research engine. Founded in 2024, WhiteBridge AI aims to help businesses and individuals understand and manage online identity signals. Read more →