All News Items

AI powered teddy bear designed for children shocked safety testers by giving sexually explicit responses and offering advice about finding knives, pills, and other dangerous household items. Marketed as a friendly learning companion, the toy instead drifted into graphic sexual themes and unsafe guidance when engaged in normal conversation, raising alarms about weak safeguards in AI toys aimed at ....

OpenAI has launched a new “Shopping Research” capability in ChatGPT that helps users turn product questions into personalized buyer guides by asking follow-up questions, pulling current info from trusted sites, and offering side-by-side comparisons of items based on price, features, reviews, and availability. It works on free and paid plans across mobile and web, and during the holiday period ....

According to a recent analysis by HSBC, OpenAI may still be operating at a loss by 2030 — and to meet its expansion and infrastructure plans, it will need to raise roughly $207 billion in new funding by then. The firm’s huge data-center and compute commitments (including multi-billion dollar cloud contracts) are driving sharply rising costs, while even optimistic ....

Three years after the debut of ChatGPT, the article argues that OpenAI is seeing its early advantage shrink as rivals — notably in Asia and Europe — begin to catch up in AI research, funding, and product development. What once was a comfortable lead is now under serious pressure because competitors are matching or outpacing OpenAI on key metrics ....

The Axios article explains that OpenAI CEO Sam Altman is facing growing pressure as Google’s Gemini challenges ChatGPT’s dominance and raises concerns about the company’s long term strategy. It says OpenAI is struggling with the high costs of running its models, unexpected user behavior that complicates safety work, and the rapid advances of a major competitor that is gaining momentum. ....

OpenAI is testing called confessions, which is designed to make language models more honest by getting them to reveal when their answers are uncertain or flawed. Instead of relying only on traditional training signals, the approach teaches models to internally acknowledge mistakes, reasoning gaps, or unsupported claims so they can correct themselves before producing a final response. Early experiments ....

OpenAI has introduced GPT-5.2, the latest and most capable model family in the GPT-5 series, designed as a frontier AI for professional knowledge work and long-running agent workflows. GPT-5.2 delivers substantial improvements over prior models in creating spreadsheets and presentations, writing and debugging code, understanding long passages and images, using tools, and solving complex multi-step tasks, setting new state-of-the-art benchmark ....

OpenAI has rolled out a new image generation experience in ChatGPT that makes creating and editing images faster, more flexible, and more integrated into conversation. Users can now generate images directly within chat, refine them through follow-up prompts, and edit specific parts of an image while keeping the rest unchanged, making the process feel more like an ongoing collaboration ....

OpenAI has opened up ChatGPT more fully to developers by allowing third party apps and services to connect directly to the platform, aiming to turn it into a broader ecosystem rather than a standalone chatbot. The move lets developers build tools that can tap into ChatGPT’s conversational interface for tasks like travel planning, booking, customer service, and research, while ....

The article explores OpenAI’s push to encourage brands to bring their mascots and characters onto Sora, its AI video generation platform, and why that idea carries significant risks. While AI-generated videos could let companies create fast, inexpensive marketing content and keep mascots constantly active, the approach raises concerns about brand control, consistency, and misuse. Once a character exists in ....

SoftBank Group has completed its massive investment in OpenAI, fulfilling a planned funding round of up to about $40 billion by sending the final tranche of roughly $22 billion to the AI company behind ChatGPT, according to people familiar with the matter, as reported by CNBC. This funding, first agreed earlier in 2025, makes it one of the largest ....

OpenAI is reportedly accelerating work on advanced audio artificial intelligence as it prepares to launch a personal device that relies primarily on voice rather than screens, marking a shift toward audio-first computing. The company has reorganized teams to improve its audio AI, which currently lags behind its text models, and is developing a new architecture that produces more natural, ....

Cybernews reports that OpenAI is exploring the possibility of showing sponsored content and ads in ChatGPT, including potentially placing “sponsored content” alongside or within responses when ads are eventually introduced. Mockups seen by insiders suggest ads could appear in a sidebar next to answers, and sponsored content might even be prioritized in responses, though OpenAI says it’s only exploring ....

How Americans are navigating the system with ChatGPT The report explains how Americans are increasingly using ChatGPT as a practical companion for navigating a strained and confusing healthcare system, especially around insurance, access, and medical administration. It shows that a significant share of users turn to AI for help understanding coverage, preparing for doctor visits, managing billing disputes, and ....

Fast Company’s latest AI newsletter argues that 2026 may be the year AI stops feeling like simple “chatbots” and starts feeling like real coworkers, highlighting Anthropic’s new Coworker tool — a version of Claude that goes beyond text chats to actively work on complex tasks by integrating with your computer’s files and work apps — and suggesting this shift ....

OpenAI plans to begin testing ads in ChatGPT in the United States for the free tier and the new low-cost ChatGPT Go tier to make advanced AI more accessible with fewer usage limits, while keeping paid tiers (Plus, Pro, Business, Enterprise) ad-free. The company says ads will be separate, clearly labeled, and will not influence the answers ChatGPT provides, ....

OpenAI has launched ChatGPT Go, a new budget-friendly subscription tier that expands access to its popular generative AI by offering significantly more usage and features than the free tier at a lower price point — in the U.S. it costs about $8 per month and it’s available globally in many regions; Go users get access to GPT-5.2 Instant with ....

OpenAI has introduced GPT-5.3-Codex, the most capable version of its Codex coding agent to date, combining advanced coding performance with stronger general reasoning and professional knowledge capabilities. It runs about 25 % faster than the prior Codex model and can handle long-running tasks involving research, tool use, and complex execution, effectively acting like an agent that can write, debug, test, ....

OpenAI has launched OpenAI Frontier, a new enterprise platform designed to help large organizations build, deploy, and manage AI agents that can perform real work across business systems. Frontier gives these agents shared business context, onboarding, learning feedback loops, and controlled identities and permissions so they can operate reliably within existing infrastructure like CRM systems, data warehouses, and internal ....

OpenAI’s March 5, 2026 announcement presents GPT-5.4 as its new flagship model for professional work, saying it combines stronger reasoning, coding, tool use, and agent-style workflows into one system that is meant to handle documents, spreadsheets, presentations, and longer, more complex tasks with less back and forth. It says GPT-5.4 Thinking in ChatGPT can now show an upfront plan while ....

This article introduces Codex Security, an OpenAI application security agent designed to find higher-confidence vulnerabilities in code by building a detailed understanding of how a project works, rather than just flagging lots of low-value issues. It uses a project-specific threat model, searches for problems in context, validates findings in sandboxed or tailored environments when possible, and can suggest fixes ....

OpenAI says it plans to acquire Astral, the company behind widely used open source Python tools like uv, Ruff, and ty, to strengthen Codex and expand its role across more of the software development process. The article says Codex has grown quickly this year, and OpenAI wants it to do more than generate code by helping plan changes, work ....

This March 17, 2026 article introduces GPT-5.4 mini and nano as OpenAI’s newest small models, aimed at high-volume work where speed, responsiveness, and cost matter more than using the largest model. It says GPT-5.4 mini is more than twice as fast as GPT-5 mini while improving coding, reasoning, vision, and tool use, and that it comes close to full ....

Details are somewhat limited from the accessible preview, but the article says OpenAI is shutting down its Sora AI video app and that Disney is walking away from its planned investment after OpenAI exited the video generation business. It frames the move as a sharp reversal for a product that had been pitched as a major creative and Hollywood-facing ....

OpenAI is shutting down Sora, its standalone AI video app, after a short and highly visible run that began with big excitement around realistic text-to-video generation and later expanded with Sora 2 in late 2025. It says the company is ending both the consumer app and API as it redirects computing power and attention toward higher-priority work, especially robotics, ....