Technology leaders struggling with an accountability gap between AI adoption and governance infrastructure
Half of UK technology leaders say the CIO (Chief Information Officer) is held personally accountable when an AI agent makes an error, more often than any other function in the business.
That’s a topline finding of Communications Reckoning: When the AI Agent Fails, Someone Has to Answer, a global survey commissioned by global business communications platform provider 8×8 Inc.
In July 2026, 8×8 commissioned independent research through Censuswide, surveying over 2,500 CIOs and CTOs (Chief Technical Officers) across the UK, USA, France, Australia, and the Republic of Ireland to ask them what AI disruption looks like from the front lines.
“What jumps out in the UK data is how many hands are already on this,” said Jamie Snaddon, EMEA MD at 8×8 Inc. “Half of the CIOs we spoke to say they personally carry the can when an AI agent gets it wrong, but one in five are also naming customer service leadership. That’s higher than anywhere else we surveyed. That’s not confusion. The fallout from AI is being felt further across the business than most people assume.”
Ownership and Responsibility
50% of UK CIOs surveyed said they are personally on the hook when an AI agent makes an error, compared with 20% for customer service leadership.
The CIO has effectively become the last line of defense for AI behaviour across the business. As AI agents take on a growing share of customer interactions, that accountability is concentrated in a single role, often without a matching increase in governance tools or authority.
Accountability also changes with company size:
- 45% of CIOs at UK organisations with 50–99 employees say they carry personal accountability, but the number is
- 62% among organisations with more than 500 employees.
- Among UK Government and public administration respondents specifically, 86% place accountability with the CIO, the highest of any sector surveyed.
Data sovereignty: a board-level concern
When asked if the location of AI sovereignty was an influence on their decision-making, 39% of UK CIOs say the location of AI infrastructure is now the primary factor in which platforms they buy, the highest figure of any market surveyed.
This follows findings showing that 91% of UK technology leaders say data sovereignty has climbed their list of priorities over the past 12 months, reflecting mounting regulatory pressure tied to GDPR and similar regulations. In France and the US, 93% report the same shift.
“UK CIOs are telling us that where their data lives now decides who gets their business,” added Snaddon. “Sovereignty is the opening question in almost every conversation we’re having with customers here.”
Vendor lock-in is a problem holding UK organisations back from consolidation
When asked about platform consolidation, UK CIOs agreed with their counterparts across the globe that cost and complexity were the main issue. However, the UK also found vendor lock-in to be a problem – more so than CIOs in the USA, France, Republic of Ireland, or Australia.
The lock-in problem is especially pronounced in the UK public sector with 71% of Government and public administration respondents citing vendor lock-in from existing contracts as their biggest obstacle – more than three times the UK-wide figure.
“The lock-in figure doesn’t surprise me,” said Snaddon. “A lot of UK organisations, especially in the public sector, are stuck in contracts signed before AI was a board-level conversation. The technology to consolidate already exists. What’s missing is a commercial path out of those agreements that doesn’t put all the migration risk on the customer.”
Governance is the biggest gap
Communications infrastructure is no longer only the supporting layer for AI. For most organisations, it has become the governance layer, the place where accountability, data control, and vendor sprawl all converge.
“The report, ‘Communications Reckoning: AI Complexity and the Multi-Vendor Data Problem,’ is available for download at https://www.8×8.com/communications-reckoning/when-ai-fails.”
This is the first of a three-part Communications Reckoning series that will continue throughout 2026 and 2027.
