BBC Halts AI Audio Pilots After Equity Threatens World Service Strike
Equity has forced the BBC to suspend trials of synthetic narration on World Service broadcasts, demanding consent rules and residual pay in...
Category: Policy
Universities Quietly Drop AI Essay Detectors After False-Flagging Dyslexic Students
Russell Group universities are abandoning AI-detection software after internal data revealed disproportionate false positives against neurodivergent and non-native writers.
ICO Fines Recruitment Firm £1.2m for AI CV-Screening That Filtered Out Carers
The ICO's first major algorithmic hiring penalty targets a tool that penalised CV gaps tied to maternity and caring duties, setting a...
DSIT Unveils £900m Sovereign Compute Cluster in Edinburgh to Cut US Cloud Reliance
The UK government has announced a £900m publicly-funded GPU cluster in Edinburgh to give startups and universities cheaper compute, but critics warn...
UK Copyright Tribunal Rules Against Stability AI in Landmark Training Data Case
A UK tribunal has ruled that Stability AI infringed copyright by training on licensed stock imagery without consent, setting binding precedent for...
Bank of England Sounds Alarm: AI Trading Bots Could Spark Synchronised Market Crashes
The BoE's Financial Policy Committee warns that similarly-trained AI trading models could herd into identical positions, amplifying flash-crash risk across markets.
OpenAI Quietly Caps API Memory Features for EU Users Over GDPR Retention Fears
OpenAI has restricted cross-session memory for European users amid GDPR scrutiny over data retention, deepening a transatlantic split in AI feature access.
Ofcom Drafts First Rules Forcing Platforms to Label AI-Generated Election Content
Ofcom's draft proposals would compel platforms to label synthetic political media ahead of local elections, raising hard questions about detection and enforcement.
AI Safety Institute’s First Mandatory Audits Expose Gulf Between Lab Claims and Reality
The UK AISI's first compulsory frontier model audits reveal stark gaps between vendor marketing and verified safety scores on autonomy and deception...