Alexandra Brodie
Partner
IP Litigation
Article
Published24 July 2026
AI agents: why governance, oversight and guardrails matter more than ever
7
minute read
Recent headlines about autonomous AI systems behaving unpredictably are a reminder that as AI becomes more autonomous, the risks become more complex.
While the specifics of any individual incident will differ, they highlight a broader challenge facing businesses: how to introduce increasingly autonomous AI systems without losing visibility, oversight and control.
AI agents can access systems, make decisions and carry out tasks with limited human intervention. The efficiencies are obvious. So too are the legal, reputational, operational and cyber security risks if appropriate guardrails are not in place.
Our recent Great Expectations report, conducted with The Financial Times, surveyed 300 senior executives from organisations with more than £500 million in annual revenue across seven international markets. It found that while 94% of executives see AI as strategically important, only 7% report strong adoption across their entire organisation, and 55% say scaling AI has been harder than expected.
The challenge is rarely the technology itself. More often, it is the governance, accountability and risk management needed to deploy AI with confidence. As organisations increasingly explore AI agents, those challenges will only become more important.
Why AI agents create new risks for your business
Generative AI tools typically respond to prompts provided by a user. AI agents are different. They can initiate actions, access systems, use tools, interact with third-party applications and complete multi-step tasks with limited human input.
1. Loss of visibility and control
One of the biggest challenges with AI agents is understanding exactly how decisions are being made. An agent may pull information from multiple sources, apply reasoning and take actions that are difficult for users to fully predict and track.
This can lead to situations where organisations struggle to explain:
- Why a decision was made
- Which data was used
- Whether appropriate policies were followed
- Who is responsible when something goes wrong
As AI systems become more autonomous, explainability becomes increasingly important for regulatory compliance, stakeholder confidence and internal accountability.
2. Data protection, cyber security and resilience risks
3. Accountability gaps
4. Regulatory and compliance exposure
5. Operational errors at scale
The businesses succeeding with AI are focusing on governance first
Organisations with stronger AI adoption are significantly more likely to have active board engagement in AI oversight and governance. They are also far more likely to involve the General Counsel at the outset of AI initiatives.
This reflects an important shift in how organisations should think about AI agents. The conversation is no longer purely about technology.
It is about:
- Risk appetite
- Decision-making authority
- Regulatory readiness
- Mitigation measures
- Governance structures
- Human oversight
Businesses that treat AI as an enterprise issue rather than a technology project are generally better positioned to scale safely and effectively.
Six practical steps to mitigate AI agent risks
1. Establish clear ownership
Every AI agent should have a named business owner responsible for:
- Oversight
- Risk management
- Performance monitoring
- Approval of significant changes
Ownership should not sit solely with IT or technology teams.
2. Define acceptable use and risk appetite
3. Implement human oversight
4. Review access permissions carefully
5. Build governance into deployment from day one
6. Test AI agents like any other critical system
AI agents are an opportunity – but only with the right guardrails
The emergence of AI agents represents the next phase of AI adoption. The potential benefits are considerable, but so are the risks.
The organisations that succeed will not necessarily be those that deploy AI agents first. They will be those that deploy them responsibly, in line with their business strategy, with clear governance, defined accountability and appropriate oversight.
As our Great Expectations research shows, the real challenge is not ambition. It is execution. Governance, accountability and oversight are what enable organisations to move AI from experimentation to enterprise-wide value.
Frequently asked questions
Q: Are AI agents riskier than generative AI tools?
A: Potentially, yes. Because they can take actions rather than simply generate outputs, AI agents often create additional governance, security, compliance and accountability risks.
Q: Do organisations need a separate governance framework for AI agents?
Q: Who should oversee AI agent deployments?
Q: What is the biggest mistake organisations make?
Q: How can businesses get started?
How prepared is your organisation?
Whether you are exploring AI agents for the first time or looking to scale existing deployments, now is the time to ensure the right governance structures, controls and decision-making processes are in place.
Our AI team advises boards, general counsel and executive leadership teams on AI governance, regulatory readiness, risk allocation, procurement, deployment and enterprise-wide adoption strategies. If you would like to discuss how to deploy AI agents safely, responsibly and at scale, please get in touch with our team.
NOT LEGAL ADVICE. Information made available on this website in any form is for information purposes only. It is not, and should not be taken as, legal advice. You should not rely on, or take or fail to take any action based upon this information. Never disregard professional legal advice or delay in seeking legal advice because of something you have read on this website. Gowling WLG professionals will be pleased to discuss resolutions to specific legal concerns you may have.
