.png)
Overview
As we approach the end of October, here’s a recap of all the new capabilities available in the Harmonic Protect portal. Check out this 3-minute video recap, or read below to learn more!
GenAI App Plan Visibility
GenAI tools offer a variety of plans to users, differing in data retention policies, data training policies, and other security features. Understanding what plans your employees are using (and putting sensitive information into) can be challenging.
This month, we introduced the ability to see all the plans in use across your enterprise, including where sensitive data has been exposed. Each of the plans has its own profile, detailing the unique risks associated.
Sensitive Data Model Expansion
Since the beginning of October, we have significantly expanded our data detection types. Our Machine Learning team has released 14 new detection types this month.
While some of these include structured data (such as access keys), the vast majority are highly unstructured datasets (such as Mergers and Acquisitions, Patent Information, and Investment Portfolio Data). We hope this will give you much deeper insights into what sensitive data is being shared into ChatGPT, Gemini, Copilot, Claude, and Perplexity.
Thank you to the teams that have worked on these models with us!
Please take a moment to look through your alerts and familiarize yourself with the data definitions. Interventions are not yet available for these new detection types, but they will be coming very soon.
Full List of New Detection Types Released in October 2024
- Access Control Policies
- Access Keys
- Billing and Payment Information
- Customer Authentication Data
- Customer PII
- Employment Records
- Financial Projections
- Investment Portfolio Data
- Mergers and Acquisitions
- Network Configuration
- Patent Information
- Sales Pipeline Data
- Security Incident Reports
- User Access Logs
Block Users Uploading Sensitive Data
One of the benefits of having highly accurate models is that we can confidently nudge the user at the point of data loss without generating excessive friction. Of course, they can often have a legitimate reason for doing so, and so providing an option to “ignore” the intervention and proceed can be helpful.
We understand that there are circumstances you may want to prevent users from uploading sensitive data entirely. We’re pleased to announce that you can now toggle these options within the “Customize Interventions” screen. Here you can choose which options a user is presented with – either “Ignore” or “Cancel”. If a user opts to cancel the submission, the will be returned to the Application without submitting the prompt.
What’s Coming Up Next?
Next month, we'll be releasing a new configuration builder designed to give you powerful, customizable options for responding to sensitive data exposures. We're excited to share more soon!
You can also expect to see even more detections, with some of those coming in line and available for interventions.
To learn more, email us at info@harmonic.security or request a demo directly here.
.png)