Jamf launches AI Governance to enhance visibility, security

Software and Apple device management company Jamf is positioning itself to help customers gain greater control over the rapidly expanding use of artificial intelligence, as businesses face the dual challenges of security and governance, as well as escalating AI costs.

Jamf recently launched AI Governance, designed to give IT and security teams greater visibility into the AI models and applications running across their Apple device fleets, allowing them to establish policies around how those tools can be used.

The product emerged from a growing need among customers to understand how, and what, AI was actually operating within their organisations, according to Jamf CEO, Beth Tschida.

"Often, executives will say yes, we're going to have AI, or maybe they don't even know they have AI," Tschida said.

By monitoring activity on endpoints, Jamf aims to provide organisations with visibility into the models being used, what those models are connecting to and what they are permitted to do.

That information can then be used to develop policies tailored to an organisation's security, operational and financial requirements.

"It allows you to make sure you are not letting AI have access to certain data that you don't want in your organisation, or using models that might be costing you more tokens than you want," Tschida said.

This approach extends Jamf's existing role in managing and securing Apple devices into the increasingly complex world of AI applications and models.

Tschida drew a parallel between AI governance and the way organisations have traditionally controlled access to software applications.

"Just like you might do with an app, you can have this one, you can't have this one. I think about it very much the same way in terms of AI governance," she said.

AI governance moves beyond IT

The rapid adoption of generative AI has created challenges that extend well beyond the IT department, with CISOs focused on data leakage, and CFOs increasingly concerned about the cost of accessing more powerful models.

The proliferation of AI models is making that challenge increasingly difficult. Companies may provide employees with access to multiple models, while newer and more capable models can carry significantly higher usage costs.

As a result, governance of AI is increasingly becoming a cross-functional issue involving executives, IT teams, security leaders and finance departments.

While many organisations have executive mandates encouraging employees to adopt AI, the challenge is finding a way to support that adoption without losing control over security or expenditure.

That means businesses will need to develop policies that balance cost, security and scalability, while also accounting for the differing priorities of executives and separate business units.

The focus on cost is also likely to evolve beyond simply measuring how many AI tokens employees consume.

In the early stages of AI adoption, some organisations treated token consumption as a measure of success, with businesses effectively tracking which employees or teams were making the greatest use of AI.

But that approach is beginning to shift towards a more fundamental question: what value is the organisation receiving from its AI spending?

"In the early days, there were some organisations who would say, we're going to have a leaderboard of how many tokens you're using," Tschida said.

"We're evolving past that to say, what's the value you're getting out of those tokens? What are the outcomes that you're getting?"

The question is likely to become more complicated as organisations increasingly use different models for different workloads and job functions.

Businesses will need to determine which models are necessary for particular tasks, whether employees require access to premium AI services and whether the additional cost is justified by improved outcomes.

Gentler approach to shadow AI

One of the evolving problems facing organisations is the growth of shadow AI, where employees use personal AI accounts or unsanctioned models to complete work.

Attempts to ban AI outright may prove ineffective, particularly as employees can access consumer AI services on personal devices and potentially enter work-related information into those platforms.

Instead, organisations should focus on enabling AI use within appropriate boundaries rather than taking a punitive approach.

Shadow AI is not necessarily driven by malicious employees. Workers may simply be looking for tools that help them perform their jobs more effectively, without fully understanding the potential security or privacy implications.

"Saying yes with the right controls is how you combat shadow AI," Tschida said.

The issue could also become more important as governments introduce new AI regulations.

While the EU AI Act came into force just over two years ago, Australia has not passed a similar act, despite Anthony Albanese recently announcing a national Office of AI.

Specific points of AI regulation will vary between jurisdictions, but companies need the ability to demonstrate what AI systems are being used and what controls have been put in place.

Jamf's approach begins with establishing visibility over AI activity before enabling organisations to produce records demonstrating their governance policies.

Opportunity for growth as AI boom continues

The AI Governance launch comes as Jamf seeks to expand the ways customers interact with its platform.

While Jamf has previously built products around a traditional user interface, now it is preparing for a future in which customers may interact with its technology through APIs, infrastructure-as-code workflows and, potentially, AI agents.

Its platform is being tweaked to accommodate customers with different levels of technical maturity, from smaller organisations that want straightforward device management through to enterprises with highly specialised workflows.

It sees APIs and other connection points as a way for customers to build their own workflows and integrations rather than relying entirely on features developed through Jamf's central product roadmap.

"If I tried to build it all on one roadmap, I put it all in our user interface, we just couldn't get to it," Tschida said.

"There are just too many different permutations of particular workflows for particular industries."

Tschida also sees the growing adoption of AI as a significant opportunity for its device management unit.

AI workloads are increasingly being developed and deployed on Macs, particularly among software developers, while organisations across sectors ranging from education to healthcare are examining how AI can be integrated into their operations.