SEND Budgets are Bankrupting Local Councils - Can AI Break the Spiral?

The Situation

England’s Special educational needs and disabilities (SEND) system is stuck in a destructive loop. Councils face rising statutory pressures and growing Dedicated Schools Grant (DSG) deficits, while families wait too long for assessment and support. As of January 2025, there were 638,700 children and young people with EHC plans (up 10.8% in a year), and timeliness deteriorated, with under half of new plans issued within 20 weeks. Explore Education Statistics+1 The Local Government Association estimates £3.15bn in cumulative high-needs deficits-projected to ~£5bn by 2025/26-and multiple councils warn of insolvency risk when accounting protections expire in March 2026. Local Government Association+1

At the same time, new evidence shows significant underdiagnosis and long waits for neurodevelopmental conditions that drive support needs. The independent ADHD Taskforce (final report, 6 Nov 2025) found consistent evidence that ADHD is under-recognised, under-diagnosed and under-treated across services-mirroring wider SEND backlogs. NHS England+1 No one wins in this set-up: councils’ finances become unstable and children go without timely support that enables learning, independence and productivity.

The Problem

SEND is information-dense: EHC plans, reports, emails, case notes, multi-agency inputs, attendance and behaviour logs-mostly unstructured, spread across systems, and costly to reconcile. Commercial tools nibble at pieces, but there’s a collective-action problem: thousands of providers and schools are too fragmented to fund a coherent national data/AI layer; local authorities cannot individually build and maintain one; and vendors have little incentive to make it truly interoperable.

The scale underscores the fragmentation: 153 English local authorities hold statutory duties; there are ~1,050 special schools and a further 1,666 mainstream settings with SEN units or resourced provisions. GOV.UK+2BESA+2 Meanwhile, the system increasingly leans on high-cost placements when early identification and consistent decision-making falter, driving budgets further out of balance. Financial Times

The Solution

There’s a gap in the market for a cutting-edge, public-interest platform: build once, use everywhere. The government should sponsor and own the core AI infrastructure for SEND, delivered as common, open standards and services that local systems and providers can plug into. Think of it as the national backbone that upgrades today’s paper-and-PDF workflow into auditable, evidence-linked decisions.

• AI-assisted documentation & search. Turn EHCPs, reports and emails into structured, searchable records with page-level citations. UK pilots show ~2 weeks/year saved per official on drafting/summarising, and an education trial showed ~31% planning-time savings with guard-railed AI-headroom that can be redeployed to direct support and quality assurance. Local Government Association+1

• Needs assessment & provision fit (NAS/PFI). A transparent, rubric-based scoring of need from EHCPs and associated evidence, with a Provision-Fit Index that flags over- or under-specification and suggests the least-restrictive, evidence-aligned options. This will raise the standard of care, enable fair, defensible negotiations on support levels and reduce avoidable overspend.

• Early-warning analytics. Cohort-level signals from attendance, behaviour, and progress help intervene sooner-supporting inclusion and reducing the need to escalate to high-cost placements and transport. GOV.UK

• Ambient scribing for casework. Draft notes, letters and review packs from practitioner interactions for human sign-off-mirroring NHS guidance now rolling out in clinical settings. NHS England

• Assurance by design. Algorithmic Transparency Recording Standard (ATRS) entries for significant models; explainability with sources; calibration dashboards for bias and inter-rater consistency; and strict access controls.

Why government sponsorship is essential

This is a textbook market failure: benefits are system-wide but costs are local; no single private actor can internalise the return; and fragmentation imposes high switching and coordination costs. Sponsorship doesn’t mean a monolith-it means government funds govern the core infrastructure and procure modular services competitively against that standard. Providers and councils adopt at low marginal cost; vendors compete on quality, not lock-in.

The ROI case is compelling. Combine conservative productivity uplifts with avoided escalation from earlier interventions, and the platform pays for itself rapidly-while improving the family journey. Local Government Association+1 Add the wider gains from getting thousands more children timely support: higher attendance, reduced exclusions, better attainment and independence; and the social return dwarfs the build cost. The Education Committee’s 2025 inquiry called out the adversarial, fragmented status quo and the need for systemic fixes; this is one. UK Parliament

What sponsorship should deliver

1. National standards & guardrails: open data model for EHCP content and outcomes; ATRS templates; privacy/security baselines; procurement catalogue aligned to these standards.

2. Core services:

   1. Document ingestion with citations

   2. Automated and accurate needs scoring

   3. Analytics layer for early-warning

   4. Note-taking and classroom support tools

   5. Audit/appeals tooling.

3. Adoption model: opt-in pilots across a representative set of local authorities and providers; shared training and MDM; incentives for vendors to certify integrations.

4. Funding approach: central capex to build the rails + time-limited opex support for onboarding; long-run unit pricing that reflects productivity savings and avoided escalation. (Harmonises with DBV/Safety Valve recovery plans rather than competing with them.) Local Government Association

Who benefits

• Families and pupils: faster assessments, clearer rationales, better-matched provision.

• Providers: less paperwork, consistent evidence, better placement stability and staff retention.

• Councils: improved timeliness and defensibility; cost control via provision-fit rather than blunt rationing.

• The UK: a flagship, mission-driven use of AI that turns a fiscal risk into a productivity win-showcasing British leadership in safe, auditable public-sector AI.

The ask

Back a government-sponsored, standards-first AI backbone for SEND. It will raise care standards, cut administrative burden, and bend the cost curve-because the right answer here is collective infrastructure, not 153 parallel reinventions.