Summary

• The Metropolitan Police has selected Neotas for an OSINT screening contract worth £6.76 million before VAT.
• Initial use covers personnel vetting, with possible expansion into firearms licensing and intelligence.
• Identity matching, context, retention, human review, and auditability will determine whether automation improves decisions.

The Metropolitan Police is preparing to automate more of its open-source and social media screening, awarding a contract that could later extend beyond personnel vetting into firearms licensing and intelligence work.

The Metropolitan Police Service has selected London risk intelligence company Neotas under a BlueLight Commercial framework. The contract is valued at £6.76 million excluding VAT, or approximately £8.1 million including VAT.

An initial three-year term covers work for MPS Professionalism and Counter Terrorism Policing Headquarters, with two optional one-year extensions. Later phases may include firearms licensing and MPS Intelligence, although the procurement notice states that neither expansion is guaranteed.

The award notice calls for comprehensive automated searching for adverse activity across material retained by original sources, using identifying information supplied by applicants. Neotas is expected to begin providing the service in August.

Vetting workload is drawing in automation

Police vetting already considers digital footprints, affiliations, public statements, and behavioural indicators. Manual work is intensive because staff must search multiple platforms, distinguish between people with similar names, assess context, and retain a defensible record of the findings.

OSINT tools can search more sources, connect related identities, translate material, and direct reviewers towards information considered potentially relevant. Used carefully, that capability may reduce repetitive work and locate material that a limited manual search would miss.

Online information is not a neutral or complete record of character, however, because accounts can be impersonated, photographs mislabelled, comments quoted without context, and social associations inferred from follows or interactions carrying little meaning.

An automated result can appear precise while resting on a weak identity match. Vetting decisions affect employment, professional reputation, access to sensitive roles, and, under a possible later phase, the ability to hold a firearms licence, so reviewers must be able to challenge a flag rather than merely approve it.

Adverse activity requires clear boundaries

The procurement language refers broadly to adverse activity over any period retained by the source. Operational policies will need to distinguish lawful expression, historic behaviour, professional misconduct, extremist affiliation, threats, and material unrelated to the decision.

Time alters the relevance of online content because a comment made many years earlier may carry a different weight from a current pattern. Deleted material may also remain in archives or commercial databases, creating questions about rehabilitation, retention, and an applicant’s opportunity to correct inaccurate information.

Automated screening engages data protection principles including accuracy, minimisation, transparency, and purpose limitation. Information being publicly accessible does not remove the responsibility attached to collecting, connecting, and using it within an official decision.

Possible future use in intelligence would materially change the purpose and risk of the system. Screening people who have applied for a role differs from analysing individuals who may not know they are being assessed, while the legal basis, oversight, and consequences may also change.

Coverage is not the same as decision quality

OSINT suppliers often compete on the breadth of sources they can search, yet wider coverage can produce more noise as well as more intelligence. Contract management should therefore examine false matches, missed material, reviewer workload, appeal outcomes, and whether the software changes the accuracy of final decisions.

Audit trails will be essential because reviewers need to know which source produced a flag, when the material was collected, how the identity match was established, what automated processing occurred, and who decided that the information was relevant.

Reliance on a proprietary platform creates additional operational questions. The Met must be able to reproduce past decisions if source access, scoring logic, or product features change, which makes exportable records and clear data return arrangements as important as conventional service levels.

BlueLight Commercial’s framework is available to police, fire, health, local government, and central government organisations, allowing the Met procurement to influence automated screening elsewhere in the public sector.

Reducing manual searches offers an obvious operational gain, but scale also increases the consequence of weak assumptions. A system that helps experienced reviewers locate and test evidence could improve vetting, whereas one that turns ambiguous online traces into unchallenged risk labels would automate the least reliable part of the process.