Independent review
AI deployment in electricity networks: independent review
Government response
No government response is recorded for this review.
About this review
Summary written by this siteIndependent review of the deployment of artificial intelligence in electricity networks, led by Lucy Yu, AI Champion for Clean Energy, for the Department for Energy Security and Net Zero. The report was published on 8 September 2026.
Recommendations 34
Recommendation 1
Government should set a strategic intent to move to full probabilistic, risk-based grid operations by 2035 and planning by 2036, and set out a roadmap for implementation as part of its upcoming AI for energy strategy
Recommendation 2
NESO should work with networks to develop probabilistic risk modelling of grid operations by 2030
Recommendation 3
Government should work with industry, Ofgem and wider society to develop appropriate risk metrics and thresholds by 2030
Recommendation 4
Government should consider a phased rollout of risk-based operation in one or more regions before a national rollout
Recommendation 5
Establish open benchmarking and leaderboard challenges for weather-dependent power system forecasting by H1 2027
Recommendation 6
NESO and DSOs should have the capability to quickly add, manage, and use large numbers of probabilistic forecasts in their core operations by 2030
Recommendation 7
NESO should consider whether setting up an energy forecasting marketplace would help accelerate AI forecasting adoption in the grid
Recommendation 8
UKRI should fund the development and integration of GPU-accelerated power flow solvers into open-source power modelling tools by Q4 2027
Recommendation 9
UKRI should focus funding on using AI to accelerate power flow models
Recommendation 10
NESO and networks should prioritise making data available that supports AI research into dynamic security assessment
Recommendation 11
NESO and networks should release an open-source high-fidelity model of a GB-like grid with an accompanying interactive training environment by 2028
Recommendation 12
Government should prioritise ensuring that benefits of AI-driven flexibility accrue to consumers
Recommendation 13
DESNZ, Ofgem and NESO should embed control systems expertise into electricity market design to account for the system dynamics created by widespread AI optimisation
Recommendation 14
Government should work with NESO and researchers to create a shared capability for system-level testing of agentic AI systems
Recommendation 15
Government should ensure AI optimisers receive effective signals of grid congestion on operational time frames
Recommendation 16
Ofgem should develop and implement new mechanisms for ensuring consumers get fair value from suppliers and asset optimisers
Recommendation 17
Government should require Ofgem and the CMA to develop and maintain active capability to prevent, detect and respond to gaming and algorithmic collusion on both short and long timescales
Recommendation 18
Government should continue to fund schemes and promote mechanisms that give fuel poor households access to flexible energy devices
Recommendation 19
Government and Ofgem should review cost recovery mechanisms for network and policy costs to ensure they remain efficient and fairly distributed as AI optimisation becomes widespread
Recommendation 20
Government should explore whether a publicly owned aggregator could deliver more value from flexibility for the public estate
Recommendation 21
Government should work with industry to agree a clear framework for defining autonomy in the grid by H1 2027
Recommendation 22
Government should work with Ofgem, NESO, industry and researchers to establish a common framework for the specification of operational design domains for grid applications by H1 2028.
Recommendation 23
Government should work with Ofgem and AISI to develop a safety case framework for autonomous AI in the grid by 2030
Recommendation 24
Ofgem should work with industry to identify an AI use case for which moving up the delegation levels is desirable, and use this to develop best practice for doing so effectively at appropriate pace
Recommendation 25
Government should develop a long-term roadmap for updating accountability and liability mechanisms to account for AI-enabled autonomy in the grid as part of its AI Strategy
Recommendation 26
Government should clearly define approval authorities and processes for deployment of AI-enabled autonomy in the grid
Recommendation 27
Government should build UK capability in defining and encoding objectives for AI systems
Recommendation 28
Ofgem should develop a roadmap for digitalising codes and regulations, including mechanisms for AI-readable compliance and automated regulatory assurance
Recommendation 29
Ofgem should redesign digitalisation incentives and funding to reinforce network-wide continuous implementation, investment, and improvement in AI, by 2028
Recommendation 30
Government and Ofgem should continue to move towards regulatory approaches that are well suited to governing the new and evolving organisational roles that AI may enable
Recommendation 31
Government should continue to prioritise, and where possible accelerate, digitalisation of the energy system
Recommendation 32
Ofgem should use its powers under the Data Best Practice Guidance to drive greater interoperability and standardisation of granularity and latency by H1 2027
Recommendation 33
Government should invest in high-resilience sovereign AI compute that can underpin future grid operations AI workloads
Recommendation 34
Government should fund a national AI for Grid Innovation Delivery hub that provides a clear pathway for delivering AI solutions across the grid
All text above is the review's.
No recommendations have this response.