Will it become harder to oversee AI systems?
A new report from the AISI Blog maps the current AI oversight landscape and identifies more than twenty pathways that could erode oversight as AI systems advance. The study draws on 25 expert interviews, literature review, and independent analysis, concluding that existing oversight methods rely on fragile properties that may degrade. It also outlines practical interventions developers can adopt now to preserve oversight and invest in emerging techniques.
The Core Development
The safety of advanced AI systems increasingly depends on the ability to oversee them: to audit models for concerning propensities before deployment, to monitor their behaviour during use, and to investigate incidents after they occur. As AI systems take on more autonomous, long-horizon tasks, the quality of oversight will determine how much we can trust them.
In a new report , we map the current landscape of AI oversight, and how it is likely to change. The report draws on 25 expert interviews across frontier AI developers, government, NGOs, and academia, together with a literature review and independent analysis of pathways to the degradation of oversight. We identify current oversight methods, and the properties of current AI systems that these methods rely on. We also examine the pathways by which they could degrade, and the technical levers available to preserve them.
Both the literature and expert opinion support the conclusion that current oversight rests on foundations that are likely to erode, and emerging methods are not yet mature enough to compensate for this erosion. However, there are practical interventions available now: developers can take steps to preserve current oversight channels and invest in emerging techniques as fallbacks for when they degrade.
Strategic Perspective & Ecosystem Impact
We then examine how these oversight surfaces might degrade moving forwards. See Figure 2 for some examples of degradation pathways:
An important part of mapping the oversight landscape was surfacing and exploring disagreements between experts. Some examples of disagreements:
Existing oversight techniques largely work well for today's AI systems, and benefits can still be gained by applying them. Developers draw on a portfolio of techniques, such as monitoring model actions, reading models' chain-of-thought reasoning, running behavioural evaluations, and probing internal activations. However, the properties of today's AI systems that make these techniques work, such as text-based reasoning, limited ability to game evaluations, and human-interpretable internal representations, are contingent and may not hold as systems are developed further.
Why it matters
Understanding how oversight may weaken is critical for maintaining trust in AI systems as they take on more autonomous, long‑horizon tasks. The report offers actionable guidance for developers and deployers to safeguard oversight.