 |
| OpenAI and ChatGPT logos are seen in this illustration taken, February 3, 2023. REUTERS/Dado Ruvic/Illustration/File Photo |
In the race to embed AI deeper into the enterprise, OpenAI’s latest move with ChatGPT isn’t just about smarter conversations—it’s about smarter control. With the rollout of Company Knowledge, ChatGPT is stepping into the heart of business operations, and that means one thing: data governance just got real.
Respecting Boundaries: Permissions First
The most reassuring part? ChatGPT honors existing company permissions. It can only access data that a user is already allowed to see—no more, no less. That’s a big win for IT leaders worried about data leakage or unauthorized access. But it also puts pressure on organizations to clean up their permission structures. If your Google Drive or SharePoint is too open, ChatGPT will reflect that vulnerability.
Admins in Control
Enterprise and Edu admins aren’t just passive gatekeepers anymore—they’re empowered architects. With the ability to manage app access and create custom roles, admins can shape how AI fits into their workflows. OpenAI also promises not to train on enterprise data by default, reinforcing its commitment to privacy. Add in encryption, SSO, SCIM, IP whitelisting, and a Compliance API for logs, and you’ve got a security stack built for serious business.
Know the Limits
Still, it’s not perfect. Users must manually activate the company knowledge feature, and when it’s on, ChatGPT loses access to web search and charting capabilities. That trade-off is temporary—OpenAI says improvements are on the way—but it’s a reminder that AI integration is a journey, not a plug-and-play solution.
Ecosystem Matters More Than Ever
The real power of ChatGPT lies in its ecosystem. With connectors rolling out for platforms like Asana, GitLab Issues, and ClickUp, OpenAI is following the playbook of IBM watsonx and SAP Joule. CIOs should take note: the value of this tool will depend heavily on how well it integrates with your existing tech stack.
From Model to Meaningful Data
This isn’t just about having a good model anymore. It’s about connecting that model to the data where work actually happens. ChatGPT’s enterprise knowledge surfacing is a bold step toward solving the AI bottleneck—making sure assistants aren’t just smart, but also informed.
What Business Leaders Should Do Now
- Audit your data permissions: CISOs and CDAOs need to ensure that access controls are tight and accurate. Use tools like Microsoft Purview or Google Workspace Admin to review sharing settings.
- Start small: Pilot ChatGPT in workflows that suffer from scattered information—like client briefings or cross-department reports.
- Set realistic expectations: Teams should understand the current limitations, especially around manual activation and web search restrictions.
- Evaluate integrations: Compare ChatGPT’s connector list with your company’s tools to gauge fit.
- Benchmark against competitors: Look at how Microsoft Copilot, Google Duet AI, and Salesforce Einstein GPT are handling AI and data integration. The real competition is in ecosystem strength.
The Bottom Line
OpenAI’s latest feature isn’t just a productivity boost—it’s a wake-up call. As generative AI moves into the private core of businesses, secure and useful data integration becomes the new frontier. For leaders, this is less about chasing the next shiny tool and more about organizing their data before someone else’s AI does it better.
