AI is transforming enterprise operations at extraordinary speed but according to discussions at Commvault’s SHIFT Melbourne event, it is also quietly rewriting the cyber risk playbook.
Hosted by Commvault, the forum brought together leaders from enterprise, government and the channel to confront a pressing question: if AI agents are becoming digital co-workers, who is securing their identities?
The answer, it seems, is not straightforward.
From Human Users to AI Agents
Research from Gartner suggests that by 2026, more than 80 per cent of enterprises will be using generative AI-enabled applications or APIs. That scale of adoption dramatically expands privileged access across systems.
Each AI agent requires credentials. Each integration creates another identity. And every identity is a potential attack vector.
Compromised credentials remain one of the leading causes of ransomware incidents globally. Meanwhile, Gartner predicts that by 2027, 40 per cent of AI data breaches will result from improper cross-border generative AI use highlighting the growing regulatory dimension of AI risk.
At SHIFT Melbourne, executives described this shift as systemic rather than incremental. AI doesn’t just add new tools it multiplies access points.
For organisations operating under frameworks such as the Security of Critical Infrastructure Act 2018, the implications are significant. Deploying thousands of AI agents could mean managing tens of thousands of new privileged identities, each with access to critical systems and sensitive datasets.
In this environment, identity becomes critical infrastructure.
The Clean Recovery Imperative
The second major theme at SHIFT Melbourne was recoverability.
Ransomware groups are no longer content to encrypt production systems they are targeting backup environments and recovery pathways. At the same time, AI initiatives rely heavily on high-integrity data. Corrupted data pipelines can undermine both operational continuity and AI performance.
Commvault used the event to demonstrate its Synthetic Recovery capability, which creates isolated, production-like environments for recovery testing and incident response. The aim is to validate recoverability without risking live data exposure effectively proving resilience before it is needed.
Through a live “Minutes to Meltdown” simulation, participants were placed inside a fast-moving ransomware crisis, testing governance structures, communication flows and technical recovery readiness.
The underlying message: resilience must be exercised, not assumed.
Channel Evolution in the AI Era
Another notable discussion centred on the changing role of partners. As organisations grapple with identity sprawl, hybrid cloud complexity and regulatory accountability, the traditional reseller model is giving way to advisory-led resilience frameworks.
With support from partners including Amazon Web Services, Hitachi Vantara and Kyndryl, SHIFT Melbourne highlighted the importance of ecosystem alignment across cloud, infrastructure and services providers.
The conversation increasingly revolves around unified resilience operations or “ResOps” integrating data protection, identity governance and clean recovery into a single operational discipline.
From Prevention to Continuity
Perhaps the clearest takeaway from SHIFT Melbourne was that cyber resilience is shifting from prevention metrics to continuity assurance.
Boards and regulators are asking a more fundamental question: if a breach occurs, can you prove you can recover cleanly and quickly?
In an AI-powered economy where automation accelerates both innovation and exposure, resilience can no longer sit in a silo. It must scale at the same pace as intelligence itself.
SHIFT Melbourne made one thing evident: in 2026 and beyond, identity protection and validated recovery are not safeguards at the edge they are the foundation of digital trust.
