In today’s evolving technology landscape, SaaS data protection remains an area of focus, particularly as organisations implement AI solutions. A recent survey highlights a gap between perceived preparedness and validated disaster recovery capabilities.
Survey highlights:
- 94% of respondents say their disaster recovery plans include agentic AI scenarios
- 32% carry out monthly testing of these plans
- 56% prioritise SaaS data protection and disaster recovery when adopting AI
- 41% have updated their disaster recovery strategies in response to increased AI adoption
- 33% of IT leaders report limited control over the use of agentic AI within their organisations
The findings indicate a difference between reported confidence in disaster recovery plans and the frequency with which those plans are tested. As AI-driven automation connects systems more closely, errors may propagate more widely, which can affect recovery processes.
The survey also points to a lag between AI adoption and updates to governance and disaster recovery approaches. A proportion of IT and security leaders report limited control over agentic AI use and express uncertainty about whether existing recovery plans fully account for these scenarios. Fewer than half of organisations report making significant changes to their disaster recovery frameworks following AI adoption.
Testing plays a key role in validating recovery processes. While regular data backups are common, comprehensive testing and validation of recovery procedures appear less frequent, particularly among smaller organisations.
The report also identifies identity and access management as an area of attention. Restoration of identity systems is tested less frequently than productivity systems, which may present operational risks, as access to other SaaS applications often depends on these systems.
Overall, the findings suggest that as AI adoption increases, organisations are continuing to adapt their disaster recovery practices to ensure they are tested, validated, and aligned with evolving operational requirements.
