OpenAI

OpenAI is an American artificial intelligence research company headquartered in San Francisco, California. Founded in December 2015 as a nonprofit by a group of technology executives including Sam Altman, Greg Brockman, Ilya Sutskever, Elon Musk, and others, OpenAI initially described its mission as ensuring that artificial general intelligence (AGI) benefits all of humanity. It has since evolved into one of the most commercially significant AI organisations in the world, credited with catalysing widespread interest in generative AI through the release of ChatGPT in November 2022.[^1]

OpenAI's research and products span large language models, text-to-image generation, text-to-video synthesis, speech recognition, and AI agent frameworks. Its work has shaped industry norms around model evaluation, safety testing, and responsible AI deployment.

Founding and Structure

OpenAI was established as a nonprofit in 2015 with initial pledges of approximately one billion US dollars from its founders and early backers. In 2019, the organisation created a "capped profit" for-profit subsidiary to attract the investment required for large-scale model training, with returns to investors limited to a multiple of their investment, with remaining returns flowing to the nonprofit. A further restructuring in 2025 converted the subsidiary into a Public Benefit Corporation (PBC), retaining 26% ownership by the nonprofit foundation. Microsoft has been OpenAI's primary cloud partner since a 2019 investment, with total committed funding exceeding 13 billion US dollars, and Azure providing the compute infrastructure behind OpenAI's training and inference operations.[^2]

Key Products and Models

GPT Series

The Generative Pre-trained Transformer (GPT) series forms the core of OpenAI's language model line. GPT-1 (2018) and GPT-2 (2019) established the paradigm of large-scale unsupervised pre-training followed by task-specific fine-tuning. GPT-3 (2020), with 175 billion parameters, demonstrated few-shot learning at a scale that had not previously been achieved, sparking broad interest in LLMs among researchers and industry. GPT-3.5, released as the backbone of the original ChatGPT, added instruction-following capability through reinforcement learning from human feedback (RLHF).

GPT-4, released in March 2023, introduced multimodal capabilities and substantially improved performance on professional benchmarks including bar exams and medical licensing tests. GPT-4o ("omni"), introduced in May 2024, unified text, image, and audio processing in a single model architecture with real-time voice capabilities. OpenAI released GPT-4.5 in February 2025 and GPT-4.1 in April 2025 as intermediate releases. GPT-5, released in August 2025, was described as a unified system integrating dynamic routing between rapid-response and extended-reasoning modes, capable of PhD-level reasoning across mathematics, coding, and science domains.[^3]

ChatGPT

Launched on 30 November 2022, ChatGPT became the fastest consumer application to reach one million users (five days) and one hundred million users (two months), establishing it as a cultural and commercial landmark. It demonstrated to a general audience that conversational AI could engage with open-ended questions, write code, draft essays, and explain complex topics, accelerating enterprise interest in generative AI across every industry.

DALL-E and Sora

DALL-E (2021) and its successors DALL-E 2 (2022) and DALL-E 3 (2023) produce high-quality images from natural-language text prompts. Sora, announced in February 2024 and released to subscribers in December 2024, is a video generation model capable of producing minute-long video clips from text descriptions, raising significant discussion about synthetic media and creative tools.

Other Products

Whisper (2022) is an open-source speech recognition model trained on 680,000 hours of multilingual audio, widely deployed as a free transcription backend. Operator (2025) is an AI agent tool that can navigate websites and complete tasks on behalf of users. The Agent Builder platform, announced at OpenAI DevDay 2025, provides a visual drag-and-drop interface for constructing agentic workflows.[^4]

Research and Safety

OpenAI publishes research across machine learning, alignment, and safety. Its Preparedness Framework assesses models for catastrophic risks including cyberweapons and biological agents before deployment. The organisation conducts red-teaming exercises and runs a bug bounty programme. It has been both a contributor to and a subject of debate about AI safety standards, with former employees raising governance concerns and regulatory bodies in several jurisdictions examining its market position.

See Also

References

References

1. Britannica. (2025). OpenAI | ChatGPT, Sam Altman, Microsoft, & History. https://www.britannica.com/money/OpenAI
2. Wikipedia. (2025). OpenAI. https://en.wikipedia.org/wiki/OpenAI
3. Script by AI. (2025). OpenAI & ChatGPT Timeline: GPT Release Dates to GPT-5.5 (2026). https://www.scriptbyai.com/timeline-of-chatgpt/
4. Wikipedia. (2025). Products and applications of OpenAI. https://en.wikipedia.org/wiki/Products_and_applications_of_OpenAI
5. Fintechnews Malaysia. (2025). The Future of Banking Is AI-First. Malaysia's CxOs Explain How That's Unfolding Now. https://fintechnews.my/55637/ai/ai-malaysia-banking-roundtable/