A UK universal bank that has put AI inside a stated cost-and-returns plan: Copilot for nearly all staff, Claude for engineers and markets operations, an FX chatbot in production, and a governance structure that names a Head of AI/ML Risk.
How is Barclays using AI?
Barclays PLC, the London-headquartered universal bank with £1,730.4bn of total assets at 30 June 2026, has made AI a named pillar of its 2026-2028 plan and has disclosed unusually specific deployments. On 10 February 2026 chief executive C.S. Venkatakrishnan said the group has more than 250 AI tools and models in use, that cloud, data platforms and AI adoption are the three pillars of its transformation, that around 90% of colleagues use Microsoft Copilot (saving more than one million hours in 2025), that GitLab for 19,000 developers lets code be implemented 15% faster, and that an FX chatbot called BARXBot delivers quotes 75% faster; the bank also said it will increase investment twofold, including embedding AI at scale. On 1 October 2026 Anthropic and Barclays announced that a Claude-powered colleague assistant serves more than 16,000 UK staff, that Claude classifies about 120,000 Global Markets emails a day, and that Claude Code is expected to reach 50% of developers by the end of 2026 and a majority of software engineers in 2027. Barclays also invested in the UK AI platform CommonAI (June 2026) and joined the FCA's AI Live Testing programme (April 2026). Its 2025 annual report on Form 20-F sets out an AI Policy with Ethical AI Principles, a Head of AI/ML Risk and a Group AI Governance Council, and a risk factor on generative and agentic AI. Its US bank, Barclays Bank Delaware, is covered separately at /banks/barclays-us.
Lead bank
Barclays Bank PLC
Total assets (group, as reported, Jun 30, 2026)
£1,730.4bn total assets[2]
Charter
UK bank holding group; Barclays Bank PLC and Barclays Bank UK PLC are PRA-authorised, Barclays Bank Ireland PLC is a significant institution under direct ECB supervision, and its US operations sit under Barclays US LLC and Barclays Bank Delaware
Headquarters
London, United Kingdom
Flagship platform
Group AI stack: Microsoft 365 Copilot, Claude and BARXBot
AI leadership
C.S. Venkatakrishnan · Craig Bright · Anne Marie Darling · Kristen Bennie · Sahana Athreya · Tony Castañon
- Scale disclosed on the February 2026 results call: more than 250 AI tools and models in use, around 90% of colleagues on Microsoft Copilot (more than one million hours saved in 2025), GitLab for 19,000 developers and a 15% faster code implementation rate.
- Claude is the newest large deployment: on 1 October 2026 Barclays and Anthropic said a Colleague Knowledge Assistant (live since 2025) has 16,000-plus UK users and over one million searches, Global Markets uses Claude on about 120,000 emails a day, and Claude Code adoption is expected to reach 50% of developers by end-2026.
- AI is written into the plan: the 2026-2028 targets (RoTE above 14% in 2028) assume investment more than doubles versus the prior three years and that AI is embedded at scale; the group also targets about £2 billion of gross efficiency savings over the next three years.
- Customer-facing AI is arriving in the US first: a conversational AI tool in the US Consumer Bank app (95% faster query responses, per the CEO) and GenAI call summaries (eight million-plus calls by February 2026). The US bank has its own page at /banks/barclays-us.
- Governance is named in the 20-F: an AI Policy defining Ethical AI Principles and prohibited, high, medium and low-risk uses, an AI Standard introduced in 2025, a Head of AI/ML Risk reporting into Model Risk Management, and a Group AI Governance Council.
- Regulators are in the loop: Barclays was selected for the FCA's second AI Live Testing cohort (April 2026), and its EU bank, Barclays Bank Ireland PLC, is directly supervised by the ECB.
- Risk language is candid: the 20-F says agentic AI with access to systems, data and tools 'creates an expanded attack surface' and names hallucination, bias and third-party model risk; the CEO called Anthropic's Mythos model a serious threat to the global banking system in April 2026.
- Ecosystem bets beyond the bank: an investment in CommonAI (9 June 2026) to build trusted, high-assurance AI infrastructure in the UK, and a Barclays Research team for data science and applied AI led from New York.
What is Group AI stack: Microsoft 365 Copilot, Claude and BARXBot?
Barclays runs AI on several named platforms rather than one. Microsoft 365 Copilot was announced for 100,000 colleagues in June 2025, integrated into the bank's own colleague productivity tool as a single agent; the chief executive told investors in February 2026 that around 90% of colleagues were on it and that it saved more than one million hours of work in 2025. In markets, BARXBot is an AI chatbot for FX quotations that he said delivers quotes 75% faster. On 1 October 2026 Anthropic and Barclays announced an expanded collaboration: Claude already powers a Colleague Knowledge Assistant used by more than 16,000 UK colleagues and sorts roughly 120,000 Global Markets emails a day, and Barclays expects Claude Code to reach 50% of its developers by the end of 2026. GitLab is rolled out to 19,000 developers (code implemented 15% faster, per the CEO), and a GenAI Centre of Excellence shares reusable components and runs hackathons. Governance sits under an AI Policy and AI Standard, with a Group AI Governance Council under the Head of AI/ML Risk.[4][6][10][9][1]
What has Barclays done on AI, and when?
- Oct 1, 2026
- Sep 14, 2026
CEO on AI and workforce training[17]
At the Barclays Global Financial Services Conference, Venkatakrishnan told Bloomberg the firm is training people for different jobs to prepare for a future in which AI increases efficiency.
- Jul 28, 2026
Half-year results: total assets £1,730.4bn[2]
The interim results announcement filed with the SEC reports total assets of £1,730.4bn at 30 June 2026, up from £1,544.2bn at 31 December 2025.
- Jun 9, 2026
Investment in CommonAI[15]
Barclays backs the UK collaborative AI engineering platform and joins its High Assurance AI programme for AI used in high-risk settings such as financial services.
- Apr 21, 2026
- Apr 17, 2026
CEO warns on Anthropic's Mythos[16]
Venkatakrishnan said in Washington that Mythos was a serious threat to the global banking system and likely to be followed by similar, more powerful cyberthreats.
- Feb 10, 2026
Annual report on Form 20-F: AI policy, council and risk factor[1]
Discloses an AI Policy and AI Standard, a Head of AI/ML Risk, a Group AI Governance Council, nine AI and ML initiatives launched in Group Finance, and a risk factor on generative and agentic AI.
- Feb 10, 2026
- Feb 5, 2026
GenAI call summaries at Barclays US Consumer Bank[20]
More than eight million customer calls summarised since October 2025; described as part of a broader global strategy to use agentic and generative AI.
- Jun 9, 2025
Where does Barclays use AI today?
| System | What it does | Status | Use case |
|---|---|---|---|
| Microsoft 365 Copilot and colleague agents[6][4] | A Colleague AI Agent, personalised semantic Content Search and a Colleague Front Door dashboard for 100,000 colleagues; around 90% of colleagues on Copilot by February 2026. | In production | Generative & agentic AI |
| Colleague Knowledge Assistant (Claude, retrieval-augmented)[10][12] | Helps Barclays UK colleagues find answers when supporting more than 20 million UK retail customers; 16,000-plus adopters, over one million searches. | In production | Generative & agentic AI |
| Global Markets email triage with Claude[10][11] | Models classify, enrich and route incoming client emails to the right operations team; about 120,000 emails a day. | In production | Generative & agentic AI |
| BARXBot FX quotation chatbot[4][9] | AI chatbot that helps colleagues initiate FX trades and delivers quotes 75% faster than the previous approach, per the CEO. | In production | Trading & capital markets |
| GitLab Duo and Claude Code for engineers[4][10][9] | GitLab rolled out to 19,000 developers; Claude Code expected to reach 50% of developers by end-2026 and a majority of software engineers in 2027. | Rolling out | AI-generated code & coding agents |
| Conversational AI in the US Consumer Bank app[4] | A conversational tool the CEO said accelerates customer query responses by 95% and enables more personalised service. | Rolling out | Customer-facing chatbots |
| GenAI contact-centre call summaries (US Consumer Bank)[20] | Reason for call, actions and next steps captured for agents; more than eight million calls summarised by February 2026. | In production | Generative & agentic AI |
| AI in wholesale credit risk systems[4] | The CEO said the bank is rebuilding wholesale credit risk architecture and using AI to aggregate and analyse data and generate reports to support fast, accurate credit decisions. | Rolling out | Credit scoring & underwriting |
| Group Finance AI and machine-learning initiatives[1] | Nine AI and ML initiatives launched in the Group Finance function in 2025, with measurable benefits reported and more in the pipeline. | In production | Generative & agentic AI |
| AI Policy, AI Standard and Group AI Governance Council[1] | Defines Ethical AI Principles and risk tiers for AI use; the Head of AI/ML Risk sets AI risk appetite and reports compliance to board risk committees. | In production | Model risk management |
| Agentic AI attack-surface risk[1][16] | The 20-F says deploying agentic AI with access to systems, data and third-party tools creates an expanded attack surface; the CEO has warned about frontier-model cyber risk. | In production | Cybersecurity |
| FCA AI Live Testing and CommonAI High Assurance programme[13][15] | Live regulatory testing of AI applications and membership of a programme for AI that makes autonomous decisions in high-risk settings. | Pilot | AI governance (general) |
Use-case links open the tracker's rules-by-use-case matrix: which documents govern each system type.
What has Barclays disclosed in numbers?
| Metric | Value | As of |
|---|---|---|
| Total assets | £1,730.4bn[2] | Jun 30, 2026 |
| AI tools and models in use across the group | More than 250[4] | Feb 10, 2026 |
| Colleagues on Microsoft Copilot | Around 90%; more than 1 million hours saved in 2025[4] | Feb 10, 2026 |
| Developers on GitLab | 19,000; code implemented 15% faster[4] | Feb 10, 2026 |
| BARXBot FX quote speed | 75% faster than the previous approach[4] | Feb 10, 2026 |
| Colleague Knowledge Assistant | 16,000+ UK colleagues; 1 million+ searches[10] | Oct 1, 2026 |
| Global Markets emails processed with Claude | About 120,000 a day[10] | Oct 1, 2026 |
| Claude Code adoption target | 50% of developers by end-2026; majority of software engineers in 2027[10] | Oct 1, 2026 |
| Applications on the cloud | 89% (75% two years earlier)[4] | Feb 10, 2026 |
| Gross efficiency savings target, next three years | About £2 billion (£1.7 billion achieved in the previous two)[4] | Feb 10, 2026 |
“Across the group, we have more than 250 AI tools and models in use.”[4]
C.S. Venkatakrishnan · Group Chief Executive · Feb 10, 2026
“AI, for example, is a transformative opportunity which contains risks that need to be managed.”[4]
C.S. Venkatakrishnan · Group Chief Executive · Feb 10, 2026
“We're moving towards AI as an increasingly agentic capability embedded within how we build, test, secure and operate technology.”[10]
Craig Bright · Group Co-Chief Operating Officer · Oct 1, 2026
“Expanding the use of Claude across Barclays is about more than adopting AI.”[10]
Anne Marie Darling · Group Co-Chief Operating Officer · Oct 1, 2026
“Artificial intelligence is transforming the way businesses operate and innovate.”[15]
Kristen Bennie · Group Head of Innovation and Partnerships · Jun 9, 2026
Who runs AI at Barclays?
Craig Bright
Group Co-Chief Operating Officer and Barclays Execution Services Co-CEO (co-leads technology, data and cybersecurity; Group Chief Information Officer from 2020)[18][10][6]
Anne Marie Darling
Group Co-Chief Operating Officer[10]
Kristen Bennie
Group Head of Innovation and Partnerships[15]
Sahana Athreya
Global Head of Data Science & Applied AI, Barclays Research[19]
Tony Castañon
Chief Operating Officer, Barclays US Consumer Bank (GenAI call summaries)[20]
What AI roles is Barclays hiring for?
10 open AI-related postings on Barclays's careers site as of Oct 5, 2026. A snapshot of current openings, not a source: postings close.
Canary Wharf, 1 Churchill Place
Canary Wharf, 1 Churchill Place
Digital Product Owner - Digital Help & AI
Canary Wharf, 1 Churchill Place (3 locations)
Knutsford, Radbroke Hall
Glasgow Campus (2 locations)
Noida, Candor TechSpace
Pune, Gera Commerzone SEZ
Director - AI Strategy & Transformation
Building 300-Whippany Campus, Jefferson Park (2 locations)
Senior Forward Deployed AI Engineer
Building 400-Whippany Campus, Jefferson Park
Liquid Financing Data/AI Engineer – VP
New York, 745 7th Avenue
Which regulators govern Barclays's AI?
| Authority | Why it matters here | Documents |
|---|---|---|
| UK (BoE / PRA / FCA) | Barclays Bank PLC is PRA-authorised; group AI model risk falls under the PRA's model-risk principles, and the FCA picked Barclays for AI Live Testing in April 2026. | PRA SS1/23 FCA FS25/5 2026 BoE/FCA AI survey |
| EU AI Act | The 20-F flags the EU AI Act as a source of new duties for AI systems, including general-purpose models, in a group whose EU clients are served mainly by Barclays Bank Ireland PLC. | Regulation (EU) 2024/1689 |
| ECB | Barclays Bank Ireland PLC is a significant institution directly supervised by the ECB, whose supervisory priorities for 2026-28 cover AI and digitalisation. | SSM supervisory priorities 2026–28 |
| Federal Reserve | Barclays US LLC is the group's US intermediate holding company; AI used by US operations falls under the Fed's model-risk guidance. The US bank is covered at /banks/barclays-us. | SR 11-7 SR 23-4 |
| FSB | A global systemically important bank adopting agentic AI sits inside the FSB's AI sound-practices consultation. | FSB AI sound practices consultation (June 2026) |
| Basel Committee | Standardising data for AI at scale ties to the BCBS 239 risk-data aggregation principles. | BCBS 239 |
What does the public record suggest about Barclays's AI strategy?
Analysis by BankingNewsAI from the sources cited on this page. Observations, not advice.
01
Several vendors, one governance spine
Copilot, Claude, GitLab and in-house tools like BARXBot are named separately, but the 20-F describes a single AI Policy and Standard overseen by one council. How tightly that spine binds third-party models is the question to watch as agentic use spreads.
02
Disclosed adoption targets are checkable
A public aim of 50% of developers on Claude Code by end-2026 and a majority of engineers in 2027 gives investors and supervisors a number to test next year, a step beyond usage anecdotes.
03
AI is tied to a cost story, not only a product story
The plan couples about £2 billion of gross efficiency savings and a doubling of investment with AI at scale. The bank has not published how much of the savings it attributes to AI, so productivity claims remain largely hours-saved and speed-up figures.
04
The risk disclosures are as specific as the use cases
Naming agentic AI's expanded attack surface and the CEO's April warning on frontier cyber models suggests the bank is treating security of agents as a gating factor for how fast it scales them.
How is Barclays using AI?
Barclays reports more than 250 AI tools and models in use (CEO, 10 February 2026). Named deployments include Microsoft 365 Copilot (around 90% of colleagues), GitLab for 19,000 developers, the BARXBot FX chatbot, a Claude-powered Colleague Knowledge Assistant for UK staff, Claude-based triage of about 120,000 Global Markets emails a day, and GenAI call summaries at its US Consumer Bank.
Does Barclays use Anthropic's Claude?
Yes. Anthropic and Barclays announced an expanded collaboration on 1 October 2026: Claude powers a Colleague Knowledge Assistant used by more than 16,000 UK colleagues, classifies and routes about 120,000 Global Markets emails a day, and Barclays expects Claude Code adoption to reach 50% of its developers by the end of 2026 and a majority of software engineers in 2027.
Who is responsible for AI risk at Barclays?
According to the 2025 annual report on Form 20-F, an AI Policy and AI Standard govern AI use. The Head of AI/ML Risk, who reports into Model Risk Management (which reports to the Group Chief Risk Officer), sets AI risk appetite and leads the Group AI Governance Council.
Is Barclays US covered separately?
Yes. Barclays Bank Delaware, the group's US consumer bank, has its own page at /banks/barclays-us, which covers its GenAI call summarisation and card business. This page covers the group.
- Barclays PLC Annual Report 2025 on Form 20-F
- Barclays PLC Interim Results Announcement 30 June 2026 (Form 6-K)
- Barclays PLC Results Announcement, year ended 31 December 2025
- BARCLAYS PLC ORD 25P (BARC.L) Q4 FY2025 earnings call transcript
- Barclays lifts profit and targets as it looks to US to boost returns
- Barclays to roll out Microsoft 365 Copilot to 100,000 colleagues
- Barclays to roll out Microsoft Co-Pilot to 100,000 staff
- Barclays rolls out Microsoft Copilot to 100,000 employees as AI adoption gathers pace (undated page, retrieved 2026-10-05)
- Scaling AI at Barclays: implementation, efficiency, and the future of customer experience (undated page, retrieved 2026-10-05)
- Barclays scales Claude to upgrade operations and improve client experience
- Barclays Expands Use of Anthropic's Claude in Efficiency Push
- Barclays ramps up use of Anthropic's Claude
- FCA announces second cohort for AI Live Testing
- Barclays, Lloyds, UBS Among Banks Picked by UK FCA for AI Tests
- Barclays backs UK AI Innovation with investment in CommonAI
- Banking industry scrambles for Anthropic's Mythos as global regulators review risks
- Bloomberg Talks: CS Venkatakrishnan
- Craig Bright (undated page, retrieved 2026-10-05)
- Barclays Research Appoints Sahana Athreya as Global Head of Data Science & Applied AI
- Barclays harnesses GenAI to deliver world-class customer service
When Barclays moves on AI, you'll read it here first.
when one of the banks we cover moves on AI, it is in the next morning's brief · six sourced stories · 7 am ET · free
plus every tracker, bank and agent page update, the morning after · leave any morning