AI agents are handling more customer interactions, which creates a new question for companies: who is responsible when an AI agent gets something wrong?
According to new research from 8×8, the answer is often the CIO.
In a survey of 2,501 CIOs and CTOs across five countries, 52% said CIOs are accountable when an AI agent fails. The problem is that the records needed to investigate those failures can be spread across separate communications systems.
For CIOs, responsibility for AI is growing faster than their ability to see what the AI is doing.
Why It Matters: AI agents are creating a new source of technology risk. Executives can be held accountable for AI decisions even when the records needed to investigate them are scattered across communications systems. Limited visibility makes AI oversight harder and puts greater weight on audit trails, data sovereignty, and infrastructure decisions.
- CIOs Are Carrying Much of the Responsibility When AI Fails: 52% of respondents say CIOs are accountable when an AI agent fails, compared with just 6% who name Legal or Compliance. The report says AI deployment has moved faster than many governance frameworks, leaving accountability with the people who have enough technical context to explain what happened. The issue cuts across company sizes. Smaller organizations may have few people equipped to own the problem, while governance processes at larger organizations may struggle to keep pace with deployment.
- AI Governance Depends on Being Able to Reconstruct What Happened: When an agent handles a customer interaction, a key record of that exchange lives in the communications platform. Fragmented systems can scatter that evidence across separate platforms, leaving gaps when an AI decision needs to be investigated. Infrastructure location is also affecting purchasing decisions. 82% of CIOs say where AI infrastructure is hosted influences platform selection, including 30% who consider it the primary factor.
- Data Sovereignty Is Adding Pressure to the Issue: 89% of technology leaders say data sovereignty has moved up their priority list during the past year. Smaller organizations can face resource constraints when several platforms create additional blind spots. Larger companies may face greater regulatory exposure. In either case, the report argues that organizations need visibility into where AI interactions and their records reside if they are going to govern those systems effectively.
- Fragmented Communications Stacks Persist for Reasons That Have Little to Do With Missing Features: Only 9% of respondents say no single platform can meet their requirements. Migration cost and complexity is the leading barrier to consolidation at 30%, followed by regulatory and data-residency constraints at 24%, vendor lock-in at 16%, and internal politics and stakeholder alignment at 15%. 8×8 argues that fragmentation can also increase compliance work by spreading records across additional jurisdictions and data policies.
- AI Is Adding Governance to the Case for Communications Consolidation: Unified communications has traditionally promised fewer vendors and simpler support. AI adds accountability because interaction records may be needed to investigate an agent’s actions. 8×8 argues that communications infrastructure can therefore play a larger role in AI governance, with unified platforms making it easier to maintain visibility across interactions and records.
Go Deeper -> Communications Reckoning: When the AI Agent Fails, Someone Has to Answer – 8×8
