UK health and care
NHS England's review of DCB0129 and DCB0160 is out for consultation until 11 September
NHS England, 2026-06-29
NHS England is consulting on revising both mandatory clinical risk management standards. AI is named as a driver: the revised standards are intended to account for artificial intelligence supporting clinical decisions, complex interconnected systems sharing data across organisations, and technologies including voice recognition and machine learning. It is described as primarily a technical consultation aimed at people with current experience of the standards.
TakeawayThis is the single biggest lever on how much assurance work your AI programme carries for the next five years, and the people who will do that work are in your team — get a response in before 11 September.
engage.england.nhs.ukCLEARvalidate offers five NHS trusts fully funded independent validation of their AVT
National CLEAR Programme, 2026-07-13
The National CLEAR Programme is funding implementation, onboarding and six months' use of a vendor-neutral assurance platform for five trusts, covering up to three clinical specialties each. It evaluates audio quality, transcription accuracy, clinical meaning and documentation quality, and returns a dashboard plus risk mitigation recommendations. Expressions of interest close 5pm on Friday 21 August; a briefing webinar runs 11 August, 12:30–13:30.
TakeawayIndependent accuracy evidence you did not pay for, at the moment HSSIB starts asking for it — if you have AVT live and no measured error rate, apply this week.
clearprogramme.org.ukManchester University NHS FT commits to 6,500 more Copilot licences a year and agents in back-office functions
HTN, 2026-08-10
MFT's board update sets out an expanded Microsoft arrangement: an additional 6,500 Copilot licences annually for three years to corporate staff, plus 1,600 frontline colleagues, with teams building AI agents to automate admin, finance and information governance tasks under human-in-the-loop protections. Alongside it, a Digital and Data Academy funded through the apprenticeship levy opens with eight courses spanning data, AI, automation, cyber security and software engineering, and a new analytics platform, ADAMS, supports work on synthetic data so teams can explore ideas without identifying patients.
TakeawayTwo transferable moves here — funding AI capability from the levy rather than a training budget, and giving teams a synthetic-data sandbox so IG review is not the first gate every idea hits.
htn.co.ukMid Cheshire Hospitals publishes a five-stage AI autonomy ladder across four domains
HTN, 2026-08-10
The trust's seven-year digital and data plan to 2033, approved by its board on 30 July, sets out AI deployment across clinical, operational, workforce and patient domains, with each domain mapped against five stages: Support (decisions made by humans only), Recommend (human in the loop), Enable (prepare or initiate, human sign-off required), Automate (execute within safe boundaries, human monitoring and override) and Autonomous (system-led, regulatory compliant). No suppliers are named.
TakeawaySteal the ladder. It gives a board a single vocabulary for "how much is this thing allowed to decide", which is the argument most AI approval meetings are actually having without saying so.
htn.co.ukCoffee break
DCB0129 and DCB0160
The two mandatory clinical risk management standards that govern health IT in the NHS. **DCB0129** binds the *manufacturer* — the supplier building a health IT system must run a clinical risk management process and produce hazard analysis and a Clinical Safety Case Report. **DCB0160** binds the *deploying organisation* — your trust must do its own clinical risk management for the system as configured, in your workflow, with your staff. They are deliberately paired: the supplier's DCB0129 output is an input to your DCB0160 assessment, never a substitute for it.
Both require a named Clinical Safety Officer, a registered clinician, who signs the safety case. The Clinical Safety Case Report is the artefact that matters: it lists the hazards considered, their controls, and the residual risk someone accepted by name.
For AI, DCB0160 is where the real work sits, because the hazards are local — how clinicians check output, what happens when it is wrong, and who notices. NHS England is consulting on revising both standards until 11 September 2026, with AI named as a driver.
