FAQ

The Mindgard Labs are a scaled-back version of our full Enterprise platform, designed to give anyone who is curious about AI Security the ability to do their own testing.

You will have full access to our AI Attack Builder system.  This lets you select a custom combination of AI attacks against a selected dataset and ML framework and then see the effectiveness of the attack in results summary report for each scenario.

Mindgard Labs is a way for us as a company to spread knowledge of how important it is to apply rigorous AI Security. We aim to educate and enable the same cyber security testing diligence that is used today on non-AI applications to be applied to AI assets.

100% free forever.  There are no costs or catches involved in signing up to our Lab environment, apart from our sign-up process.  Any communications from which, you can fully opt out of.

The Mindgard AI Security Labs allow each registered user the ability to run up to 5 AI attack combinations and then analyse the results once complete.

Absolutely! Each AI attack that a user builds is run in our cloud environment and GPU datacenter against the exact parameters chosen. This allows us to add more attacks, models, datasets and frameworks in the future.

Given that we have curated the AI attack scenarios in the AI Security Labs, each attack scenario can involve everything from simple vision models to hosting our own LLM, and hence each AI attack can vary considerably in duration. A typical AI security test takes several minutes to complete, subject due to number of users actively using the platform.

Furthermore, our AI Security Labs has finite resources, so please bear with us as we process each user's attack submission.

In the coming weeks, we will also release a notification system to update you when your AI attack results are available in the AI Security Labs portal.

The Mindgard Enterprise version can be deployed within a container or on 'bare metal'. We work with all our Enterprise customers to recommend the ideal deployment scenario within their own environments.

The Enterprise version of the Mindgard platform makes every feature of the platform available. This ranges from our entire range of AI attacks (largest commercially available globally) through to access to our 'AI threat intel' approaching a million AI attack scenario results, which in turn drives our remediation recommendation engine.

The Enterprise Mindgard platform is truly flexible and can be deployed in cloud, on-premise and even within an air-gapped environment.