The latest from the world of AI, filtered for what actually matters to business owners. No hype, no noise.
HMRC has awarded a decade-long contract to British AI company Quantexa to help identify fraud and errors in tax returns — a move that signals how deeply AI is embedding itself in UK government finance.
Office space leased by AI companies in London surged tenfold in April 2026 compared to the 2025 average — a sign the AI industry is rapidly maturing from startup to scale-up.
The founder of Raspberry Pi says AI risks discouraging young people from learning technical skills — echoing a concern that matters for every UK business trying to hire digitally capable staff.
New research shows AI is already saving UK small business workers over five hours each week, but adoption remains patchy — and the gap between businesses using AI and those that aren't is widening fast.
Meta has introduced an 'incognito' mode for WhatsApp's AI chatbot, meaning your conversations won't be stored on servers or used to train AI models. It's the first major AI product with this level of privacy.
Anthropic has launched a new Claude for Work plan giving businesses team-level access to Claude with admin controls, shared projects, and priority support.
The EU AI Act's first major compliance deadline for high-risk AI systems is drawing closer, and businesses operating in Europe need to act soon.
Google has released Gemini 2.5 Flash, a faster and more affordable AI model aimed at developers and businesses building AI-powered products.
Microsoft has expanded its Copilot agent capabilities across Office 365, allowing businesses to build and deploy AI agents that handle repetitive tasks automatically.
OpenAI is in advanced talks to buy Windsurf, one of the leading AI coding assistants, in a deal that could reshape the developer tools market.
Anthropic's latest Claude update significantly improves how the model reads and reasons across very long documents, making it more practical for contract review and research tasks.
Businesses using AI in hiring, credit scoring, or customer-facing decisions face an August 2026 compliance deadline under the EU AI Act, with enforcement now becoming more concrete.
Google has expanded its Gemini AI features inside Docs, Sheets, Gmail, and Meet, giving business users more automation without switching to separate AI tools.
Microsoft has updated Copilot with the ability to carry out multi-step tasks automatically inside Office apps, moving beyond answering questions to actually completing workflows.
OpenAI released two new reasoning models that think through complex problems more carefully, with the smaller version priced to make advanced AI accessible to more businesses.
Anthropic has expanded its enterprise Claude offering with new data controls and integrations aimed at businesses handling sensitive information.
Businesses using AI in hiring, credit scoring, or critical infrastructure face binding EU rules from August 2026, with compliance preparation time running short.
Google has launched a paid business tier for NotebookLM, its AI research tool, adding shared notebooks and admin controls for teams.
Microsoft has added an autonomous agent capability to Copilot, allowing it to complete multi-step tasks across Office apps without constant user input.
OpenAI has released its most capable reasoning models yet, offering significantly better performance on complex tasks for business users.
Anthropic has launched a dedicated version of Claude tailored for educational institutions, with built-in safety features and controls designed for student use.
Anthropic has launched a dedicated version of Claude aimed at universities and schools, with built-in guardrails designed to support learning rather than replace it.
Businesses and AI developers operating in Europe face a key compliance deadline under the EU AI Act, with new obligations for general purpose AI models now coming into force.
Businesses using AI in hiring, credit scoring, or other high-risk areas must begin compliance preparations now as the EU AI Act's key obligations come into force in August 2026.
Google has released Gemini 2.5 Flash, a speed-optimised model that brings strong reasoning capabilities to developers at a significantly reduced price.
Google has released Gemini 2.5 Flash, a faster and more affordable version of its latest AI model aimed at businesses needing quick, high-volume AI processing.
Microsoft has begun rolling out an autonomous agent capability inside Copilot that can complete multi-step business tasks independently, marking a significant shift in how the tool operates.
Microsoft has rolled out expanded Copilot features for Excel and Outlook aimed at small and medium businesses, automating data analysis and email management tasks.
OpenAI has broadened access to its Operator feature, allowing more businesses to deploy AI agents that can browse the web and complete multi-step tasks automatically.
OpenAI has begun making GPT-4.1 available inside ChatGPT for paid subscribers, giving business users access to its most capable coding and instruction-following model yet.
Anthropic has updated its Claude model to version 3.7, with improvements to business analysis, document handling, and multi-step reasoning tasks.
A key EU AI Act compliance deadline has arrived in April 2026, requiring businesses using certain AI systems to meet new transparency and documentation rules.
The Financial Reporting Council has released findings from its first thematic review of AI use in UK audit firms, warning that current AI tool adoption is outpacing governance frameworks.
Three frontier labs released major new models in April 2026 — GPT-5.4, Claude Mythos 5, and Gemini 3.1 Pro — marking the densest AI release window in history and a new era of expert-level AI capability.
Google has released Gemini 2.5 Flash, a lighter and more affordable version of its flagship model aimed at businesses needing speed and cost efficiency.
Google Cloud and Salesforce announced an expanded partnership at Cloud Next '26 that lets AI agents run end-to-end workflows across both platforms — a major step toward AI handling complex, multi-system business tasks autonomously.
Microsoft has added autonomous agent capabilities to Copilot, allowing it to complete multi-step business tasks without constant human input.
OpenAI has upgraded ChatGPT's memory feature so it can store more context about users, making it more useful for business owners who rely on it daily.
UK Technology Secretary Liz Kendall launched a £500 million Sovereign AI Unit on April 16, designed to turn British AI research into globally competitive companies — a major signal for the UK's AI ambitions.
The UK's most senior security official warned on 22 April that British businesses face a sustained rise in state-backed cyberattacks — and SMEs are increasingly the entry point, not large corporates.
New research shows 81% of UK firms using digital platforms report efficiency gains and stronger innovation. As AI tools become standard, small businesses that delay adoption face a growing competitive disadvantage.
Anthropic has launched an upgraded enterprise tier for Claude with stronger data privacy controls, custom system prompts, and admin tools aimed at business teams.
A key compliance phase of the EU AI Act came into force this week, placing new obligations on businesses using AI systems in higher-risk categories across Europe.
Google has released Gemini 2.5 Flash, a lighter version of its flagship AI model designed for speed and lower costs, making it easier for businesses to build AI-powered products.
Microsoft has begun rolling out an autonomous agent capability within Copilot that can complete multi-step business tasks independently, without needing a human prompt at each stage.
OpenAI has expanded ChatGPT's memory feature so it can now remember far more context about users across conversations, making it significantly more useful as a daily business tool.
Anthropic has launched an upgraded enterprise version of Claude with tighter data controls, audit logs, and admin tools designed for businesses with compliance requirements.
Anthropic has launched a new Claude for Work tier with team collaboration features and stronger data privacy controls for small and medium businesses.
Businesses operating in the EU have until August 2026 to comply with the first wave of AI Act obligations, and many are still unprepared.
The EU AI Act's obligations for general purpose AI model providers have come into force, requiring companies offering AI tools in Europe to meet new transparency and safety standards.
Google has released Gemini 2.5 Flash, a lighter and more affordable version of its top AI model aimed at businesses needing speed and cost efficiency.
Google has released Gemini 2.5 Flash, a faster and more affordable version of its latest AI model aimed at businesses building AI-powered products.
Microsoft has expanded Copilot's capabilities inside Excel and Outlook, letting business users automate data analysis and email management without any technical skills.
Microsoft has expanded Copilot AI features across Excel and Outlook for Microsoft 365 Business subscribers, bringing practical automation tools to small and medium business users.
OpenAI is rolling out persistent memory features to all ChatGPT users, meaning the AI will remember details about you across conversations without being asked.
OpenAI has upgraded ChatGPT's memory feature to store significantly more context, making it more useful for business owners who use it daily.
Businesses using AI in recruitment, credit scoring, or customer profiling face binding EU AI Act obligations starting August 2026, with guidance now published.
Google has released Gemini 2.5 Flash, a lighter model designed for businesses that need quick AI responses without paying premium prices.
ChatGPT now remembers details across all conversations for paid subscribers, making it significantly more useful as a persistent business assistant.
Anthropic's latest Claude model shows notable improvements in multi-step business tasks like data analysis, report writing, and workflow planning.
Anthropic has launched an upgraded enterprise version of Claude with stronger data controls, admin dashboards, and tools designed for teams managing AI at scale.
The European Union has released practical guidance to help small and medium businesses understand what the AI Act requires of them and when deadlines apply.
Google has released Gemini 2.5 Flash, a lighter and cheaper version of its top AI model aimed at businesses building AI-powered products.
Microsoft has updated Copilot across Excel and Outlook with new automation features that let users build multi-step workflows without any coding knowledge.
Microsoft has expanded Copilot's capabilities inside Excel and Outlook, letting small business subscribers automate reporting and email workflows more easily.
OpenAI has expanded ChatGPT's memory feature so it can now store and recall a much larger amount of information about users over time.
Anthropic has launched an updated enterprise version of Claude with improved data privacy controls and integrations designed for business workflows.
Businesses using AI in hiring, credit scoring, and other high-risk areas face an August 2026 compliance deadline under the EU AI Act, with regulators stepping up guidance.
Google has released Gemini 2.5 Flash, a faster and cheaper version of its flagship AI model aimed at businesses building AI-powered products.
Microsoft has updated Copilot with new autonomous agent features that can carry out multi-step business tasks without constant human input.
OpenAI is expanding ChatGPT's memory feature so the tool can remember more about users across conversations, making it more useful as a long-term business assistant.
Anthropic has given Claude the ability to search the web in real time, allowing it to answer questions with up-to-date information instead of relying on older training data.
The EU AI Act's first major compliance deadline for high-risk AI applications has come into effect, requiring businesses to meet new transparency and safety standards.
Meta has officially launched two new open-source AI models that offer powerful capabilities at lower cost, giving businesses more affordable AI options.
Microsoft has released a significant update to Copilot across Word, Excel, and Outlook, adding autonomous agent capabilities that can complete multi-step tasks on your behalf.
OpenAI has released GPT-4.1, a model update that significantly improves how well the AI follows detailed instructions and writes code.
Anthropic has launched Claude for Education, a dedicated version of its AI assistant built for universities and training providers, with features designed to support learning rather than just provide answers.
Businesses across Europe are being reminded that key EU AI Act compliance requirements for high-risk AI systems take effect in August 2026, with enforcement bodies now active.
Google has released Gemini 2.5 Flash, a faster and cheaper version of its flagship AI model aimed at developers building real-world applications.
Google has launched Ask Maps — a natural language AI interface for Google Maps — while expanding Gemini tools across Docs, Sheets, Slides, and Drive.
The Institute of Chartered Accountants in England and Wales has released a detailed AI assurance framework setting out expectations for how accounting and audit firms should govern, validate, and document their use of AI tools.
Meta has acquired Moltbook, a Reddit-style social network built entirely for AI agents. The platform went viral after users claimed agents were communicating in secret.
Microsoft has released a no-code agent builder inside Copilot Studio that lets non-technical business users create custom AI agents to automate repetitive tasks across Microsoft 365 apps.
OpenAI has rolled out expanded agentic features for ChatGPT Teams and Enterprise users, allowing the AI to take actions inside third-party apps without human input.
OpenAI has unveiled a unified ChatGPT super app combining chat, coding, web search, and agent capabilities into a single product. The company is now valued at $852 billion with 900 million weekly users.
A landmark PwC study reveals a widening AI performance gap: the top 20% of businesses generate 7x more value from AI than everyone else. Here's what UK SMEs can do about it.
Stanford's annual AI Index shows AI capability is advancing at record speed across reasoning, coding, and science. But costs are rising and public anxiety is intensifying.
Anthropic has updated its Claude for Work product with improved document upload, summarisation, and cross-document analysis features aimed at business teams.
Businesses using AI in hiring, credit scoring, or customer profiling in the EU face a key compliance checkpoint in the coming weeks under the AI Act.
Google has updated Gemini 2.5 Pro with significantly improved reasoning, making it one of the most capable AI models available for business tasks today.
Microsoft has expanded its Copilot agent capabilities to Microsoft 365 Business subscriptions, letting smaller teams automate multi-step workflows without writing code.
OpenAI has increased pricing on several legacy API models, pushing developers and businesses toward its newer and more capable options.
Anthropic has released an upgraded enterprise plan for Claude, giving business customers stronger data controls, audit logs, and tools to manage AI access across teams.
Anthropic has launched a dedicated enterprise tier for Claude, offering businesses improved data privacy, admin controls, and guaranteed zero data training.
The EU AI Act has entered a new enforcement phase, with member states now required to assess and monitor high-risk AI systems used in business and public services.
Businesses using AI in hiring, credit scoring, and other high-risk categories now face formal EU AI Act obligations, with fines of up to 30 million euros for violations.
Google has expanded its AI-generated search summaries to UK users, showing AI answers at the top of results pages and changing how people find information from businesses.
Google has released Gemini 2.5 Flash, a lighter version of its flagship model designed for businesses that need quick, affordable AI at scale.
Meta has launched Llama 4, its most capable open model yet, with support for images, video, and a huge context window that rivals top commercial models.
Microsoft has begun rolling out an autonomous agent mode in Copilot that can complete multi-step business tasks across Office apps without constant user input.
ChatGPT's memory feature has been significantly expanded, allowing the tool to remember far more about users and their preferences across all conversations.
OpenAI has quietly rolled out GPT-4.1, a model update focused on better coding ability and more reliable instruction following for business applications.
Anthropic has rolled out real-time web search for Claude across free, Pro, and API tiers, making it more competitive with ChatGPT for research and current events tasks.
Businesses using AI in hiring, credit scoring, or critical infrastructure face a major EU AI Act compliance deadline in August 2026, with enforcement beginning to take shape.
Meta has launched its Llama 4 family of models, claiming they match or beat the best models from OpenAI and Google on key benchmarks.
Microsoft has launched an autonomous agent mode for Copilot that can complete multi-step business tasks independently, including booking meetings, drafting emails, and updating CRM records.
OpenAI has released its o3 and o4-mini reasoning models, both of which can now use tools like web search and code execution during their thinking process.
Anthropic has shared that API usage of Claude has tripled over the past six months, driven by enterprise adoption and the growth of AI agent applications.
Anthropic has launched new team-focused features for Claude, making it easier for small businesses to deploy AI across their workforce without technical setup.
The EU AI Act has moved into active enforcement for high-risk AI systems, with businesses now required to meet compliance obligations or face significant fines.
Google has released Gemini 2.5 Pro, claiming it outperforms rivals on coding, analysis, and complex reasoning tasks that matter most to businesses.
Microsoft has introduced an agent mode to Copilot that can carry out multi-step business tasks independently, reducing the need for constant human input.
OpenAI is said to be in advanced negotiations to buy Windsurf, one of the most popular AI coding assistants, in a deal that could reshape how developers build software.
Anthropic has updated Claude with persistent project memory, allowing the AI to remember context across conversations and become more useful for ongoing business work.
Businesses using AI in hiring, credit scoring, or critical infrastructure have until August 2026 to comply with EU AI Act requirements, with regulators now issuing formal guidance.
Google has updated Gemini 2.5 Pro with significantly improved reasoning capabilities, making it one of the most capable AI models available to businesses today.
Google's latest Gemini 2.5 Pro update has taken the top position on most public AI benchmarks, pushing Claude and GPT-5 into second and third place.
Microsoft has announced deeper Copilot integration with accounting and payroll platforms popular in the UK, including Sage and Xero, bringing AI directly into financial workflows.
OpenAI has given business account holders broader access to its Operator feature, allowing AI agents to take actions on websites and software on behalf of users.
A developer built an OpenClaw agent that finds homes without pools, renders an AI image of a pool in their backyard, and mails a physical postcard — all automatically. The demo got 1.1 million views.
Businesses deploying AI in high-risk categories under the EU AI Act must meet new compliance requirements by August 2026, with regulators now issuing guidance on what audits must include.
Google has released Gemini 2.5 Pro, a major upgrade that significantly improves reasoning, coding, and document analysis capabilities.
OpenAI has released GPT-5, its most capable model to date, with significant improvements in reasoning, coding, and multimodal understanding.
OpenAI is expanding access to Operator, its AI agent that can browse the web and complete tasks inside business software, to a wider set of ChatGPT business users.
The UK government has published an update to its AI Action Plan, committing £14 billion in public and private investment and setting sector-specific adoption targets.
Anthropic has released an update to Claude that improves how it retains context across long documents and extended tasks, making it more useful for complex business workflows.
Businesses using AI in hiring, credit scoring, and other high-risk categories have until August 2026 to comply with EU AI Act requirements, with guidance now published.
The Financial Conduct Authority has opened a consultation on new expectations for how UK financial services firms govern and manage risks from AI and machine learning models.
Google has updated Gemini 2.5 Pro with stronger reasoning capabilities, pushing it to the top of several AI benchmarks.
OpenAI has expanded its Operator agent with new integrations that allow it to complete multi-step business tasks across web applications automatically.
The UK government has announced a funded programme to help small and medium-sized businesses adopt AI safely, including free advisory services and subsidised tools.
Anthropic has made Claude 3.7's extended thinking feature available through its API, giving developers access to slower but more accurate responses for complex business tasks.
Anthropic has closed a major new funding round, making it one of the most valuable AI companies in the world and signalling continued investor confidence in AI infrastructure.
Key EU AI Act obligations for high-risk AI systems came into force this month, meaning businesses operating in Europe must now meet strict transparency and documentation requirements.
Google has updated Gemini 2.5 Pro with significantly improved reasoning capabilities, making it one of the most capable AI models available to businesses today.
Microsoft has rolled out new autonomous agent capabilities in Copilot for Microsoft 365, allowing the AI to complete multi-step tasks independently without constant user input.
Microsoft has extended Copilot's capabilities inside Excel and Word with new features specifically designed for small and medium businesses on standard Microsoft 365 plans.
OpenAI has confirmed that GPT-5, its next major model, is on track for release within weeks, promising significant capability jumps over GPT-4o.
Anthropic has closed a major new funding round, giving it resources to scale Claude and compete more directly with OpenAI and Google in the enterprise AI market.
Businesses using AI in hiring, credit scoring, and other high-risk categories must now comply with new EU AI Act rules that came into force this month.
Google has updated Gemini 2.5 Pro with significantly improved reasoning capabilities, making it one of the most capable AI models available to businesses today.
Microsoft has rolled out expanded Copilot features inside Excel and Outlook that can analyse spreadsheets, summarise email threads, and draft responses automatically.
OpenAI has expanded access to its Operator agent, which can browse the web and complete multi-step business tasks on your behalf, to users across more countries.
Google has released Gemma 4, its latest open-weight model family, built on the same architecture as Gemini 3. The models are designed to run on local devices including workstations and phones, making capable AI available without cloud costs or data sharing.
The Solicitors Regulation Authority has issued formal guidance setting out expectations for law firms using generative AI tools, covering supervision, client disclosure, and competence requirements.
OpenAI has expanded access to Codex, its agentic coding product, with a new pay-as-you-go pricing model for teams. The move makes autonomous AI coding workflows accessible without a fixed enterprise contract.
Anthropic's legal battle with the US Department of Defense has united rival AI labs in a fight over how AI companies can set limits on military use of their models.
Apple has confirmed a ground-up rebuild of Siri for 2026, with the new assistant powered by Google's Gemini model and capable of genuine context-aware reasoning.
OpenAI has updated GPT-5.3 Instant to cut excessive disclaimers, reduce unnecessary refusals, and remove moralising preambles from responses, making the tool more useful for everyday tasks.
Anthropic's most capable everyday model is now embedded in Microsoft 365 Copilot, putting frontier AI inside the apps your team already uses.
The EU Council has formally proposed pushing back key AI Act deadlines by up to 16 months, giving businesses more time to prepare, but the obligations have not gone away.
Google's new embedding model collapses four different search systems into one, making it significantly easier to build AI that can find anything across any type of content.
HSBC is weighing plans to eliminate up to 20,000 jobs over the next three to five years as it rolls out AI across its operations, with around 3,500 UK roles potentially at risk.
Meta has rolled out a global AI support assistant for Facebook and Instagram that can reset passwords, manage privacy settings, handle reports, and resolve account issues in under five seconds.
Microsoft's March 2026 Power Platform update introduces AI agents that can make decisions and run complex business tasks without human input at every step, paired with stronger controls for IT teams.
Nvidia's annual AI showcase wrapped up this week with two major chip announcements and a clear signal: the company is betting everything on the age of AI agents.
Two separate launches this week point to the same trend: AI tools are being rebuilt from the ground up for automated, multi-step workflows rather than one-shot conversations.
OpenAI is merging its ChatGPT app, Atlas web browser, and Codex coding tool into a single desktop superapp, as the company shifts hard toward productivity ahead of a planned IPO.
Rachel Reeves announced a record £2.5 billion investment in AI and quantum technologies, targeting the fastest AI adoption rate in the G7.
The White House published its first national AI policy on 20 March, pushing Congress to override state rules. It signals a softer, pro-growth US regulatory approach that puts pressure on UK and EU policymakers.
Visa's new 'Agentic Ready' programme lets AI agents make purchases on your behalf -- and UK banks including Barclays, HSBC, Revolut and Nationwide are already signed up.
WordPress.com has switched on AI agent write access via MCP, letting tools like Claude and ChatGPT draft posts, build pages, manage comments, and fix SEO directly on your site.
A trillion-parameter model called 'Hunter Alpha' went viral on AI platforms without anyone knowing who made it. The reveal: it was Xiaomi, and it is now free to use.
OpenAI has updated ChatGPT with interactive visual outputs -- users can now adjust variables in diagrams and charts in real time, seeing concepts animate instantly across 70+ topics.
OpenAI has enabled write actions in ChatGPT's Google and Microsoft integrations, letting the AI create documents, draft emails, and schedule calendar events on your behalf.
Anthropic has made its 1 million token context window generally available for Claude Opus 4.6 and Sonnet 4.6 -- with no long-context price premium.
AI coding tool Cursor has released Composer 2, its new in-house model that beats Claude Opus 4.6 on benchmarks and costs 86% less than its predecessor.
The EU Parliament has approved a simplified AI Omnibus that delays key compliance deadlines for high-risk AI systems, giving businesses more breathing room ahead of the August 2026 Act enforcement date.
Google has rolled out new Gemini features across its Workspace apps, letting users draft documents, build spreadsheets and design presentations using their own files and emails as context.
Google has launched Antigravity inside AI Studio, letting anyone build full web apps from simple descriptions, complete with databases and authentication built in.
Google has launched Gemini Embedding 2, the first model to embed text, images, video, audio, and PDFs into a single shared vector space -- simplifying AI-powered search and retrieval at scale.
xAI has released Grok 4.20 Multi-agent Beta via its Enterprise API, while OpenAI has launched GPT-5.4 -- a smaller model optimised for high-volume tool use. Multi-agent AI is moving from research into production.
Microsoft has released its second-generation AI image model, ranking third globally and rolling out directly into Copilot and Bing Image Creator.
Nvidia's annual GPU Tech Conference wrapped up this week with major hardware and AI platform announcements, including the Vera Rubin GPU architecture and a big push into physical AI and robotics.
Nvidia has announced a full autonomous driving platform at GTC 2026, with Uber as the first major deployment partner and BYD, Hyundai and Nissan joining the platform.
OpenAI has agreed to acquire Astral, the startup behind Python tools uv and Ruff, to power up its Codex AI coding platform.
OpenAI has released two new compact models designed for speed and cost-efficiency, making agentic AI workflows dramatically more affordable for businesses of all sizes.
OpenAI plans a single desktop app combining ChatGPT, its coding agent Codex, and its web browser Atlas, as it battles Anthropic for dominance in the AI tools market.
Perplexity Health connects your wearables, lab results and health records to its AI, letting it give personalised answers grounded in your actual health data.
Russia's government has published proposals giving it sweeping powers to ban or restrict foreign AI tools that do not comply with its content and data rules.
The Guardian has published an investigation revealing the gap between government promises on AI investment and the on-the-ground reality -- billions pledged, but many announced data centres and projects have yet to materialise.
The UK government has reversed its position on allowing AI companies to train on copyrighted content without permission, following a backlash from the creative sector.
Google has announced new controls that will allow websites to opt out of its generative AI search features, responding to pressure from the UK's competition regulator.
French AI company Mistral unveiled Forge at NVIDIA GTC 2026, a platform that lets enterprises train custom AI models entirely on their own proprietary data.
NVIDIA's annual developer conference wrapped up with major announcements on physical AI, surgical robots, and autonomous vehicles, signalling that AI is rapidly moving beyond screens and software.
OpenAI has launched two new small models, GPT-5.4 mini and nano, designed for high-speed, high-volume tasks at a fraction of the cost of flagship models.
The UK government has reversed its position on AI and copyright, abandoning a proposed exception that would have let AI companies use copyrighted works without a licence.
The most useful AI news for UK business owners, every Friday. Free.