· Venture Capital Tracker · investment-strategies · 2 min read
AI Score’s $5.4M Seed: Runtime Governance for Enterprise Agents
Fuel Ventures led a $5.4M seed for London’s AI Score — NCSC-rooted founders selling live oversight of generative and agentic AI as boards lose visibility into what agents actually do.
VCT data record
Funding event facts
Source-backed financing and transaction details. Unknown terms remain undisclosed rather than estimated.
AI Score raises $5.4M seed
London AI governance startup AI Score closed a $5.4M Fuel Ventures-led seed with GALLOS returning to scale runtime controls for generative and agentic AI.
- Event type
- Funding Round
- Event date
- Sep 3, 2026
- Stage / label
- Seed
- Amount
- $5.4M
- Confidence
- Company Disclosed
Company / target: AI Score
Lead: Fuel Ventures
Participants: GALLOS Technologies
Sources: tech.eu
AI Score raised $5.4 million in seed funding led by Fuel Ventures, with founding investor GALLOS Technologies returning — coverage dated September 3–4, 2026 (Tech.eu; Sky News). Angels named in coverage include MMC co-founder Alan Morgan and Ventura Capital’s Mo El Husseiny. Valuation was not disclosed. Prior ~$1 million pre-seed (Nov 2025) brings disclosed capital to roughly $6.4 million.
Spine: enterprises deployed chatbots with policy PDFs; agents that can act broke that model. Boards need a live watchtower, not another annual AI ethics workshop.
Key facts
| Field | Detail |
|---|---|
| Company | AI Score |
| Round | $5.4M seed (~£4M) |
| Lead | Fuel Ventures |
| Also in | GALLOS; named angels from finance/security |
| Founders | Alex Harland, Benita Tibb (+ Kewley) |
| Advisor (reported) | Sir Jeremy Fleming (ex-GCHQ) |
Who uses the product — and for what job
Users: risk, security, legal, and transformation teams in regulated and brand-sensitive orgs. Company describes demand from law firms, FTSE 250 companies, a UK fintech, and global consumer brands — logos not published in the sources used here.
Job: continuous visibility into where generative and agentic AI runs, who owns it, what it can do, and how to constrain it without freezing adoption.
Why now
- EU AI Act enforcement steps in 2026 raised documentation and oversight pressure for companies operating in Europe.
- Agent tooling (tool use, browsing, write access) turned “shadow IT chat” into operational risk.
- National-security alumni packaging governance as software is the talent signal buyers recognize.
Why Fuel Ventures / GALLOS — portfolio fit
Fuel Ventures has leaned into vertical AI software; GALLOS is a security/defence/resilience studio that incubated the company. That pairing is distribution into regulated UK buyers plus product DNA from people who built government cyber capability — not a generic SaaS seed.
Fuel Ventures and GALLOS have no /fund/ pages here. Compare adjacent AI security coverage: HiddenLayer’s $100M Series B and AIR’s agent supply-chain seed.
Competitive map
| Approach | Tradeoff |
|---|---|
| Policy + annual audit | Cheap; blind to runtime agent behavior |
| Cloud CSP AI controls | Native to one stack; weak cross-SaaS view |
| AI security / red-team platforms | Strong attack focus; different buyer motion |
| AI Score governance layer | Org-wide visibility pitch; early proof |
What is not proven
- Valuation, ARR, and growth rate (company says H1 2026 revenue “soared” without numbers).
- Named customers and retention.
- Differentiation vs CSP-native controls once agents standardize.
Practical takeaway
- Founders (AI governance): Sell runtime ownership, not another policy generator.
- Investors: Diligence integration breadth and false-positive load on security teams.
- Operators: Relevant if agents can already spend money, change tickets, or email customers unsupervised.
Sources
Follow Venture Capital Tracker in Google
Add VCT as a preferred source to make our venture-capital coverage easier to find in Google Search.
Editorial note: AI tools assisted with research, structure, or drafting. Venture Capital Tracker retains human editorial responsibility for factual accuracy, relevance, and source quality before publication.
