OpenAI has launched ChatGPT Enterprise, a business version of its generative-AI service that offers unlimited access to higher-speed GPT-4, a 32,000-token context window, advanced data analysis and enterprise security controls while promising that customer prompts and company data will not be used to train OpenAI models. The company’s Aug. 28 announcement positions the product as a direct answer to a central problem in corporate AI adoption: employees want to use generative tools, but organizations are reluctant to expose sensitive information through consumer services.
OpenAI says teams associated with more than 80% of Fortune 500 companies have adopted ChatGPT since its public launch nine months ago, based on accounts registered with corporate email domains. ChatGPT Enterprise is intended to turn that bottom-up experimentation into a centrally managed business service with stronger privacy, security and deployment controls.
The enterprise product removes GPT-4 usage caps
The most visible change is access. ChatGPT Enterprise removes the usage limits applied to other ChatGPT plans and offers GPT-4 performance up to twice as fast, according to OpenAI. It also expands the context window to 32,000 tokens, allowing users to work with substantially longer documents, files and conversations.
OpenAI’s release notes confirm that Enterprise includes unlimited access to Advanced Data Analysis, the feature previously called Code Interpreter. That capability allows ChatGPT to execute calculations and analyze uploaded data, bringing spreadsheet, statistical and data-cleaning workflows into the same conversational interface used for writing and research.
The underlying model is GPT-4, which OpenAI introduced in March as a more capable successor to GPT-3.5. The company’s original GPT-4 research release described improved performance on professional and academic benchmarks while also emphasizing continuing limitations, including hallucinations and reasoning failures. Enterprise changes how companies can deploy that model; it does not eliminate those underlying model risks.
Privacy and governance are the core business proposition
OpenAI says ChatGPT Enterprise data are encrypted at rest using AES-256 and in transit using TLS 1.2 or later. The service is SOC 2 compliant and includes single sign-on, domain verification, an administrative console, bulk member management and an analytics dashboard. Most consequentially for corporate risk teams, OpenAI says customer prompts and company data are excluded from model training by default.
A contemporaneous TechCrunch report notes that those safeguards address concerns that have led some companies to restrict employee use of consumer generative-AI tools. Businesses have been especially cautious about workers pasting source code, contracts, financial data or unreleased product information into systems whose data-handling rules may not satisfy internal security policies.
That makes ChatGPT Enterprise as much a governance product as a model-access product. Central administrators can determine who has access, verify company domains and observe usage at an organizational level. The aim is to move AI use from unmanaged individual accounts into a controlled environment that can be evaluated by security, legal and compliance teams.
Microsoft is pursuing the same enterprise opening
OpenAI is not entering an empty market. Microsoft, its largest strategic partner, introduced Bing Chat Enterprise in July with commercial data protection and said prompts and responses would not be used to train underlying models. Microsoft’s Inspire announcement made the service available in preview to eligible Microsoft 365 customers and outlined a future standalone price of $5 per user per month.
A separate Microsoft corporate announcement also set a $30-per-user monthly price for Microsoft 365 Copilot when broadly available. That product is designed to reason across business data in Microsoft Graph and applications such as Word, Excel and Teams, creating a different kind of integration than the standalone ChatGPT interface.
The parallel launches illustrate how quickly generative AI is moving into a competition over enterprise distribution. OpenAI controls the ChatGPT brand and the direct customer interface; Microsoft controls a large installed base of workplace software and cloud identities. Both are emphasizing that commercial adoption requires data protection and administrative control in addition to model capability.
Enterprise AI still requires human verification
ChatGPT Enterprise’s security controls address how data are handled, not whether model outputs are always correct. GPT-4 can still generate inaccurate statements, misread ambiguous instructions or produce plausible but unsupported conclusions. Businesses using it for research, coding, data analysis or drafting therefore need review processes proportionate to the consequences of an error.
That distinction is important because enterprise deployment can magnify both benefits and mistakes. A single employee using ChatGPT experimentally creates limited organizational exposure; a company embedding the tool across hundreds or thousands of employees can generate large productivity gains but also scale flawed workflows if outputs are accepted without verification.
OpenAI is initially selling Enterprise through its sales organization rather than publishing a standard per-user price. Early customers named by the company include Block, Canva, Carlyle, Estée Lauder, PwC and Zapier. The company says it plans smaller-team offerings and additional customization capabilities in the future.
The broader shift is already visible. Generative AI began 2023 largely as a consumer and developer phenomenon. By late summer, the competitive focus is moving toward managed workplace deployment, where security, privacy, identity and data governance matter alongside raw model performance. ChatGPT Enterprise is OpenAI’s clearest effort yet to make GPT-4 a standard corporate tool rather than an application employees use around the edges of official information systems.
