Third-party cyber evaluations involving OpenAI models
openai.com·14h ago
Cloud & Infrastructure
The New Stack·21d ago·5 min read
The enterprise strikes back: Keep your data, ditch the AI bill.
TL;DR
As enterprises increasingly adopt AI, concerns about data security are intensifying. Companies are now prioritizing the protection of their data over the costs associated with AI usage.
✦ Why It Matters
Engineers should implement data governance frameworks today to ensure compliance and protect sensitive information in AI projects.
Key Takeaways
- Evaluate the risks of using proprietary AI models for your data.
- Investigate open-source AI alternatives for internal use.
- Monitor AI labs' data usage policies closely.
- Consider developing in-house AI models to reduce dependency.
Full Summary
As the AI cost crisis subsides, enterprises are becoming more concerned about the security of their data when using proprietary AI models from companies like OpenAI and Anthropic. These enterprises not only face high costs for AI usage but also risk feeding their data back into the AI labs, which then sell improved models back to them.
This creates a dependency cycle that raises alarms, especially as AI labs expand into software categories dominated by their clients. Microsoft, for instance, is at a critical juncture, potentially needing to explore open-source models or develop its own.
While major AI labs claim they do not use API data for training, skepticism remains, prompting calls for more secure, private AI solutions. The conversation is shifting towards creating open models that can be trained on internal data, allowing enterprises to maintain control over their information.