Policy
-
Financial watchdog puts human oversight back into AI
The Financial Stability Board has finalised 12 voluntary practices for AI in finance, covering board accountability, model evaluation, human oversight, cyber risk and dependence on technology suppliers.
-
EU cyber programme makes room for AI security
The EU cybersecurity work programme has been amended to support implementation of its AI cybersecurity action plan. The change establishes a programme priority but does not announce additional funding or completed projects.
-
Microsoft faces a new UK competition test
Britain’s competition regulator has proposed designating Microsoft’s business software ecosystem with strategic market status. Copilot competition, security product bundling, interoperability and cloud licensing are among the issues under scrutiny.
-
Met Police ordered to tighten third-party data controls
The Information Commission has issued a reprimand and enforcement notice after Metropolitan Police personnel used an unapproved crime intelligence platform to share personal data. The force must improve its oversight of external technology.
-
Salesforce wins signed £2bn HMRC CRM contract
HMRC’s £2 billion CRM contract with Salesforce has been confirmed through official procurement records. The agreement covers an initial ten years, with extensions potentially taking it to 15, and includes planned AI-assisted casework.
-
Finland brings AI into classified government workflows
Finland’s Foreign Ministry has begun piloting UMAI, an AI assistant for certain classified and confidential material. The deployment will inform the development of a secure platform for wider central government use.
-
TikTok faces another obstacle in gatekeeper appeal
An adviser to Europe’s highest court has recommended rejecting ByteDance’s appeal against TikTok’s Digital Markets Act gatekeeper designation, clarifying the evidence needed to challenge a regulatory classification.
-
Fraud refunds come before negligence disputes
Europe’s highest court has ruled that payment providers must immediately reimburse unauthorised transactions even where customers may have been grossly negligent. Banks can pursue recovery separately.
-
Palantir settlement returns Met software deal to competition
Palantir and London’s policing authority have settled their legal dispute over a blocked £50 million Metropolitan Police software contract, clarifying the grounds for rejection while leaving a new procurement process to determine the eventual supplier.
-
DBS brings AI into safeguarding evidence reviews
The Disclosure and Barring Service has disclosed an AI system that summarises safeguarding referrals for caseworkers, with source-linked evidence, mandatory human review and restrictions on automated decision-making.
