AI News — Page 68
The latest AI developments shaping UK business and technology
AISI: GPT-5.5 matches Anthropic Mythos on offensive cyber tasks
UK AI Safety Institute says GPT-5.5 hit 71.4% on Expert-tier cyber tasks and completed its 32-step corporate-network attack range — second model to do so after Mythos.
System C ambient-AI tool deployed across 15 councils, lifting assessment capacity 40%
System C's FormFlow Assistant is in 15 English council adult-social-care teams and reports 68% documentation efficiency gain and 40% end-to-end assessment uplift.
47% of UK job seekers have had an AI interview; 30% walked away from one
Greenhouse survey of 1,132 UK candidates shows AI interviews now standard at the top of the funnel — and a third of UK applicants have abandoned a hiring process because of one.
Made Smarter ships Cheshire AI toolkit with Scan-Pilot-Scale framework for SME factories
Government-backed Made Smarter North West and DBT's AI Champion publish a SME manufacturing AI toolkit centred on a Scan-Pilot-Scale adoption pathway.
Harvard trial: AI beats ER doctors on text-based triage; 31% of UK doctors now use AI weekly
Harvard study published in Science: OpenAI's o1 reached 67% diagnostic accuracy on triage text vs 50-55% for human pairs. UK uptake is widening alongside the trial data.
ASOS partners with Microsoft on AI Stylist, agents now write 15% of code
ASOS deploys conversational AI Stylist, Azure Datalake and Microsoft 365 Copilot. Agents handle 50% of customer enquiries; 93 agentic use cases identified.
EU AI Act trilogue stalls overnight — August compliance date back in play
Parliament and Council failed to agree on the Digital Omnibus reform after 12 hours of talks. Without a deal next month, Annex III rules apply on 2 August 2026.
UHSussex AI assistant cuts NHS patient call waits from 30 minutes to under three
University Hospitals Sussex deployed Netcall's conversational AI assistant. Call wait times down 90%, abandonment down 75%, 1,500 queries a day handled digitally.
UK financial services lacks shared AI governance standard, Zango report warns
Lloyds, Monzo and Stripe leaders warn that UK FS firms are solving AI governance independently. US and Singapore have sector-specific standards; the UK does not.
Claude Mythos Preview solves 30% of human-unsolvable bioinformatics tasks
Anthropic's BioMysteryBench benchmark pits Claude against expert virologists on 99 real-world bioinformatics questions. Top model solves 30% of tasks no human could.