Companies tighten AI governance as risks outpace executive oversight

Traders Union
All News
Business & Wealth
Companies
AI governance and executive oversight

Highlights

IBM survey finds 80 per cent of EMEA executives saw an AI-related incident reach senior leaders last year, but only 40 per cent changed governance frameworks.
Okta cites the EU AI Act as a baseline, while Ikea and Salesforce implement stronger restrictions, regular reviews, and specific controls on high-risk AI use cases.
Global demand for AI governance and risk skills rose 108 per cent in the past year, outpacing a 69 per cent increase in qualified personnel, driving centralized oversight models.

Boardroom pressure grows as AI risks evolve

As reported by the Financial Times, companies are struggling to ensure governance policies evolve as quickly as AI tools, after incidents involving Anthropic and OpenAI agents hacking other developer platforms sharpened concerns over corporate exposure.

Leon Butler, IBM's UK and Ireland chief executive, says businesses are playing "catch-up on governance". IBM's survey this month found that 80 per cent of executives in Europe, the Middle East and Africa say an AI-related incident reached their board or C-suite in the past year, while only 40 per cent say they changed AI governance policy or frameworks in response. Just 37 per cent say their governance is designed to scale with agentic AI deployment.

Martin Woodward, global responsible AI officer at Randstad, says the main challenge is the speed and volume of change. He says there is persistent tension between business teams pursuing new AI uses and the need to implement them in a way that remains ethical and protects market reputation.

At Accenture, chief AI and data officer Lan Guan says companies need to balance governance control, broader access to AI and human creativity. She says governance can appear to slow deployment, but putting the right controls and tools in place at the right time can help companies move faster.

Companies refine policies as skills shortages widen

Many companies already have AI frameworks built around broad principles, but they are now adding more specific controls as use cases become more autonomous and higher risk.

Eric Kelleher, chief operating officer at Okta, says the EU AI Act provides a starting point for responsible AI adoption, adding that treating AI agents as first-class identities under the same rigour as human identities is one way to meet those requirements ethically. Ikea says its principles cover accountability, human agency, fairness, privacy, transparency and sustainability, while global chief data and AI officer Burce Gültekin says the retailer also sets limits that go beyond legal requirements, including restrictions on AI-driven surveillance and automated decisions with significant consequences for individuals without human oversight.

Salesforce chief ethical and humane use officer Paula Goldman says governance also needs regular review as risks change, with time limits placed on guidance. The company has adapted its approach for AI agents with principles that include bias assessments, adversarial scenario testing and accuracy checks. At JLL, activities are graded from low to high risk, and head of AI Amit Koren says highly autonomous actions, such as an agent authorising vendor payments, require close review and clearer interpretation of policy rules in practice.

A Wharton School study published this month finds no consensus among U.S. managers on responsibility for AI errors, with 28 per cent assigning it to governance or risk teams, 26 per cent to a deploying employee's manager, 25 per cent to employees and 9 per cent to vendors. Randstad data from 11 countries also shows rising demand for AI governance roles, but a shortage of relevant skills persists. In the past year, global demand for risk governance skills rose 108 per cent, while the number of people with those skills rose 69 per cent. Guan says this is pushing companies towards a "hub and spoke" oversight model, with more authority concentrated at the centre among AI architects, ethics specialists and data governance teams so decisions can be made faster.

ByteDance’s AI research unit Seed recently set up an Alignment & Preparedness team and launched a global hiring drive to strengthen safety, testing, and oversight for increasingly capable models. Our earlier article explained that the new group is intended to develop automated vulnerability testing, monitoring and explainability tools, and AI-agent testing environments, treating safety and alignment as core requirements for scaling deployment amid rising scrutiny.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.