The world is in the midst of an unprecedented era of artificial intelligence innovation. Looking ahead, there will be two types of companies: those that lead the way in AI and those that risk irrelevance.
For organizations that take AI seriously, the composition of their workforce will change dramatically.
Today, their workforce is purely human. Tomorrow, it will expand exponentially to include a variety of AI workers—including apps, agents, robots, and even humanoids. We will live in a world where connected AI agents and humans work together to orchestrate all sorts of complex workflows. And I believe this will translate into a massive increase in productivity and capacity with significant shared benefits.
Imagine what a human population of 8 billion people can do if we have a combined capacity of 80 billion.
But the question is, how can we make this transition safely and securely?
The adoption of artificial intelligence brings new risks
Ensuring the safety and security of artificial intelligence in the enterprise is a difficult and relatively new problem. This is because AI applications are built differently and add a new layer to the stack: models. Unlike traditional applications, AI models can behave unpredictably, and the reality is that most organizations will use multiple models across public and private clouds. This multi-model, multi-cloud and multi-agent environment requires a new approach to safety and security.
When models fail, there can be serious consequences. Security issues—such as bias, toxicity, or inappropriate output—must be addressed alongside threats from external actors who exploit vulnerabilities to steal confidential data or otherwise compromise your security. Model vendors and app developers will implement their own safeguards, but these safeguards, while necessary, will inevitably be fragmented and insufficient.
Ultimately, your security teams will need a common layer of visibility and control. Not only do they need to see and understand where AI is being used in your organization (both by users and application developers), but they also need to constantly validate and enforce your preferred controls over how AI models, applications, and agents behave.
Introducing AI Defense: Rethinking Safety and Security for AI
You need to move fast with AI, but you absolutely cannot afford to sacrifice safety and security for speed. That’s why we announced it today at our AI Summit Cisco AI Defense– a solution designed to eliminate this trade-off and allow you to innovate fearlessly.
AI Defense provides robust protection in two critical areas:
- Access to AI applications: Third-party AI applications can increase productivity, but they pose risks such as data leakage or malicious downloads. With AI Defense, you get full visibility into app usage and enforce policies that ensure safe and secure access—all with support Cisco Secure Access and enhanced with AI-specific protection.
- Building and running an AI application: Developers need the freedom to innovate without worrying about vulnerabilities or security issues in their AI models. AI Defense discovers your AI footprint, validates models to identify vulnerabilities, applies guardrails and enforces them in real-time across public and private clouds.
AI Defense is built on two game-changing innovations we pioneered: continuous AI validation and protection at scale.
Validation at scale
You need to make sure your AI models are fit for purpose and free of vulnerabilities, unexpected behavior, data poisoning or other issues.
With traditional apps, you would use a “red team” of people to try to break the app and find vulnerabilities. Unfortunately, this is not realistic for non-deterministic AI models.
This is where our AI Algorithmic Red Team capability comes into play. It’s one of the main reasons Cisco acquired Robust Intelligence last summer. They are a team of AI security pioneers who have developed what we believe is the world’s first algorithmic red team solution.
The AI Algorithmic Red Team sends a succession of rapid variations to the model to try to get it to give answers it shouldn’t. Instead of a red team of thousands of people trying to break a model out of jail for weeks, we’ll do it in seconds.
It’s like playing a game of 100 questions. But because it’s automated, it’s a 1 trillion question game. And thanks to artificial intelligence, 1 trillion looks small.
Once the AI defense finds weak spots, it will recommend a guardrail for you to use. And it does it continuously. So whenever your model changes or whenever a new threat type appears, your model will be revalidated and the updated railings will be applied.
Protection at scale
With our platform approach, we can protect AI in a way that only Cisco can deliver.
We already connect traditional security directly to the network. You get thousands of enforcement points, everywhere you need them, distributed, close to users and close to workloads. These checkpoints can be located in an application in a public cloud, in infrastructure in a private cloud, on a server, on a top-of-rack switch, or even at the edge.
Artificial Defense takes full advantage of this platform approach, so your AI guardrails are also hyper-distributed and available wherever you need them. You get complete visibility into your entire AI footprint and control you can apply everywhere.
Importantly, AI Defense is also seamless for developers. In fact, it is invisible. There are no agents, no libraries required, nothing to slow down development. This means you can move quickly to create new AI experiences and innovate for your customers.
Purpose-built technology backed by unrivaled intelligence
AI Defense is built on purpose-built technology and our own AI models powered by Scale AI. By working closely with leaders like Scale AI and leveraging our own proprietary intelligence, AI Defense provides unmatched insight and ensures fast, efficient and accurate protection.
Unlocking the full potential of AI
I’m incredibly proud of what our team has accomplished with Cisco AI Defense. This solution enables organizations to move quickly, innovate boldly and unlock the full potential of AI – securely and without compromise.
Learn more about Cisco AI Defense and how it can protect your AI journey:
Read: Cisco AI Defense: Comprehensive Security for Enterprise AI Adoption
Watch the video
Register to play the AI Summit online
https://www.ciscoaisummit.comMore information
Share:
