ChatGPT: Everything you need to know about the AI-powered chatbot

You might pretty easily catch that you shouldn’t put glue on your pizza to make the cheese stick. Others are more subtle, like when a model creates fake citations and attributes them to real authors. While agentic AI definitely marks a turning point in human-computer interaction, moving from tool use to collaboration, the next step is integrating these agents and actually deriving value.

OpenAI to introduce a program to grow AI infrastructure

AI can also surpass human scale, even open up new applications that were not previously possible. For instance, makeup company L’Oreal launched a virtual beauty advisor to scale that exclusive service beyond their retail locations, reinventing the way they think about interacting with their customers at scale. I began as a cognitive psychologist studying language—the science underpinning today’s large language models (LLMs)—and later led teams at YouTube, Spotify, and Instagram.

The chatbot provides recommendations, images, and reviews of products in various categories such as fashion, beauty, home goods, and electronics. OpenAI has released two new reasoning models, o3 and o4 mini, just two days after launching GPT-4.1. The company claims o3 is the most advanced reasoning model it has developed, while o4-mini is said to provide a balance of price, speed, and performance. The new models stand out from previous reasoning models because they can use ChatGPT features like web browsing, coding, and image processing and generation. But they hallucinate more than several of OpenAI’s previous models.

The Broken Mental Model

LLMs remix oceans of language to mirror those signals with unsettling precision. Even when we know we’re talking to code, our nervous systems often can’t (or won’t) care. A recent Harvard Business Review analysis shows that “therapy and companionship” is the number one generative-AI use case, drawing nearly a third of all activity.

How good is ChatGPT at writing code?

When an LLM doesn’t find the exact best answer for your question, it’ll give you one that sounds correct but isn’t. Or it could make something up out of thin air that appears plausible. Some of these errors, called hallucinations, are pretty obvious.

What controversies have surrounded ChatGPT?

AI won’t tell you when you’re losing touch with yourself. Consider Character.AI, where users chat with custom personas. The site now logs roughly 200 million visits a month, and the average session lasts about 29 minutes—longer than the typical Instagram scroll. These aren’t drive-by queries; they’re erotic, confessional, intimate exchanges. The kind we used to reserve for those who actually know us. But OpenAI is involved in at least one lawsuit that has implications for AI systems trained on publicly available data, which would touch on ChatGPT.

OpeanAI intends to release its “first” open language model since GPT-2 “in the coming months.” The company plans to host developer events to gather feedback and eventually showcase prototypes of the model. The first developer event is to be held in San Francisco, with sessions to follow in Europe and Asia. OpenAI has launched three members of the GPT-4.1 model — GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano — with a specific focus on coding capabilities. In the competition to develop advanced programming models, GPT-4.1 will rival AI models such as Google’s Gemini 2.5 Pro, Anthropic’s Claude 3.7 Sonnet, and DeepSeek’s upgraded V3.

For instance, an analyst could use the company’s slide deck and documents to develop an investment thesis. OpenAI made a notable change to its content moderation policies after the success of its new image generator in ChatGPT, which went viral for being able to create Studio Ghibli-style images. The company has updated its policies to allow ChatGPT to generate images of public figures, hateful symbols, and racial features when requested. OpenAI had previously declined such prompts due to the potential controversy or harm they may cause. However, the company has now “evolved” its approach, as stated in a blog post published by Joanne Jang, the lead for OpenAI’s model behavior.

• Over the weekend, users on social media criticized the new model for making ChatGPT too validating and agreeable.
• Another important step is ramping up scarcity — that means taking away resources so employees need to do more with less.
• “It gives you this authoritative, confident answer, and that is the danger in these things,” said Alex Mahadevan, director of the MediaWise media literacy program at the Poynter Institute.
• OpenAI for Countries is part of efforts to support the company’s expansion of its AI data center Project Stargate to new locations outside the U.S., per Bloomberg.
• Before generative AI, the key was to judge the source, Mahadevan said.

Referrals from ChatGPT to news publishers are increasing. Since Google launched its AI Overviews in May 2024, the percentage of news searches that don’t lead to clicks on news websites has increased from 56% to nearly 69% by May 2025. ChatGPT, OpenAI’s text-generating AI chatbot, has taken the world by storm since its launch in November 2022. What started as a tool to supercharge productivity through writing essays and code with short text prompts has evolved into a behemoth with 300 million weekly active users.

Epoch AI, the research institute behind FrontierMath, discovered that o3 achieved a score of approximately 10%, which was significantly lower than OpenAI’s top-reported score. After introducing a data residency program in Europe in February, OpenAI has now launched a similar program in Asian countries including India, Japan, Singapore, and South Korea. The new program will be accessible to users of ChatGPT Enterprise, ChatGPT Edu, and API.