In the first post of our series, we discuss trends related to a more aggressive rollout and adoption of GenAI. In 2024, things were straightforward: “AI App” usually meant a chatbot, chat happened in a browser, and interacting with AI required some effort on the part of the user. Looking ahead, all of these concepts are evolving rapidly.

AI Everywhere – Last year, the concept of an AI app vs a regular app was clear, so blocking access by domain was possible. Today, with AI embedded in many commonly used applications, this approach won’t work. Security teams are working out strategies for controlling access to and risks from acceptable apps that happen to have generative AI components embedded in them. Sorting out data security with AI is inseparable from business processes within applications will be a challenge for enterprises in 2025.

Adoption of Copilot – Business leaders are driving adoption of Copilot use for writing, spreadsheets, etc. while developers are using Github Copilot to improve coding productivity – the benefits are significant. However, anecdotes such as “someone typed in a prompt and saw all of the CEO’s email” or “someone typed in a prompt and saw everyone’s salary” are common. In 2025, companies will face challenges balancing the benefits of Copilots with the privacy risks, as these tools may access data stored in internal systems without considering access privileges.

Dive Deeper: Unlocking the Full Potential of Microsoft Copilot with WitnessAI

Advent of Apple Intelligence – Apple Intelligence continues to expand its feature set, and with over 153 million iPhone users in the U.S., each new feature is adopted on a massive scale. New features like AI summaries for emails, messages, websites, and more have already been launched. Even phone calls may have an AI summary created with a button press. With these features used by employees of all kinds there is significant exposure for business. Even disabling these features en masse for employees by leveraging an MDM does not disable them for employee’s families who may still leverage them in the home, and on their phones. This is all to say nothing of the fact that Apple Intelligence AI is also present on MacBooks, and Home Assistants as well.

Configuring AI for Security and Enablement – AI is clearly a powerful tool for analysis en masse. This provides an opportunity to leverage AI to better understand employee behaviors, and to provide training, and guidance to both improve job performance, but also to lower exposure, and risk. For example, AI technologies can find, and train employees on the front end of data loss instead of relying solely on the backstop of DLP technologies.

Part 2: AI Gets Complicated

Part 3: AI Gets Compliant