Keeping (security) up with the pace of innovation in AI: GPT-4o & ChatGPT for Desktop

Prompt Team
May 28, 2024

Every week, more than a hundred million people use ChatGPT. Now, many of them will be able to use it on their desktops too. Prompt Security’s Agent ensures employees can safely and securely use the desktop app of ChatGPT.

A couple of weeks ago, OpenAI did it again and made the headlines after announcing the release of GPT-4o (“o” for “omni”), a new flagship multimodality model that can reason across audio, vision, and text in real time. According to the company, GPT-4o is their latest step in pushing the boundaries of deep learning, this time in the direction of practical usability.

To put things in context, prior to the release of GPT-4o, one could use Voice Mode to talk to ChatGPT but with a pipeline of three models and the resulting latencies: 

  1. one simple model transcribes audio to text, 
  2. GPT-3.5 or GPT-4 takes in text and outputs text, 
  3. a third simple model converts that text back to audio. 

This process means that the main source of intelligence, GPT-4, loses a lot of information: it can’t directly observe tone, multiple speakers, or background noises, and it can’t output laughter, singing, or express emotion.

With GPT-4o, OpenAI has trained a single new model that combines all the modalities end-to-end across text, vision, and audio, meaning that all inputs and outputs are processed by the same neural network. 

We’ve been so fascinated by all of this that we couldn’t help but bring it up in this blog post. But today we’re here to talk about another announcement that came out at the same time as the release of GPT-4o: 

OpenAI launched ChatGPT for Desktop, available for macOS and soon for Windows too

For both free and paid users, OpenAI has launched a new ChatGPT desktop app for macOS that is designed to integrate seamlessly into anything a user is doing on their computer, and ask questions to ChatGPT with a simple keyboard shortcut. One can also take and discuss screenshots directly in the app. ChatGPT can respond based on what it “sees.”

ChatGPT Desktop app. Source: OpenAI

For desktop-based applications, Prompt Security delivers a lightweight GenAI Security agent

While the primary usage of ChatGPT and similar tools has been happening primarily on browsers, the reality is that desktop applications with GenAI features like copilots or assistants of some kind have been around for a while already and are central in the day to day operations of employees, whether it’s Zoom, Slack, Grammarly, etc. 

One of our core use cases at Prompt Security is securing the use of GenAI tools by employees - regardless of where they’re using them. For desktop applications that are based on GenAI or include some GenAI components, we have a lightweight agent that deploys in minutes and gives the same functionality as our browser extension for Shadow AI detection, data privacy and governance. 

We are thrilled to announce that Prompt Security now extends its support to the ChatGPT desktop app. With a simple procedure involving just a few clicks, you can deploy the agent to your employees' macOS computers, resulting in immediate, comprehensive protection.

You need an adaptive ally for AI security

The pace of innovation in AI is unprecedented. While its rapid advancement captivates us and technologists will continue pushing the boundaries of its potential, keeping up with the new security, ethical, and legal risks it introduces will be a challenge, as this wave of innovation opens the door to a whole new array of risks.

Security needs to keep up with the speed of AI innovation, and at Prompt Security we’re on a mission to allow organizations to fully unleash the power it brings with it.

Want to learn more? Book time with us.

Share this post