Good morning from the wAIve team in London.

Today, Monday 26 October 2026: A day for big spending and awkward refunds. The Chancellor is putting hundreds of millions into a "distinctively British" approach to AI in Wednesday's Budget, while Western Australia wants its money back from Deloitte after AI hallucinations crept into a report on merging three universities. Meanwhile, OpenAI is offering discounts of up to 90% to tempt Anthropic's customers away, so somebody is getting a bargain this week.

  POLICY  

UK Chancellor John Healey has said he'll use Wednesday's Budget to commit hundreds of millions of pounds to AI, describing the approach as "distinctively British." The announcement frames AI as a national security and economic competitiveness priority, explicitly responding to what Healey called an accelerating global technology arms race.

THE PLAY: Substantial public funding tends to unlock private co-investment and procurement opportunities, particularly in defence, healthcare, and infrastructure. This one comes with a warning; budget leaks can often be wrong.

  RESEARCH  

An economist at Apollo Global Management has published a stark finding: despite trillions invested globally, AI has delivered zero measurable financial payoff to corporate America outside the tech sector.

THE PLAY: Measurable ROI is still the holy grail. The risk for boards is that investment without return becomes untenable, leading to cuts at precisely the moment when returns might be about to happen. Trust your decisions. Hold your nerve. And be willing to pivot as the landscape changes.

  BREAKING  

The Western Australian government is demanding a refund from Deloitte, after discovering that a feasibility report on merging three universities contained AI-generated hallucinations and a cost discrepancy of over two hundred and sixty million dollars. The government paid $160k for the study. Deloitte has not disputed the errors.

THE PLAY: Consulting firms, law firms, accountancies, and any organisation producing reports for clients or regulators: check your professional indemnity.

OpenAI is offering discounts of up to ninety percent to enterprise customers currently using Anthropic's Claude, according to The Decoder. Companies including Rippling, Replit, and DataCamp report cutting AI costs significantly after switching. Anthropic's growth has slowed as a result.

Why it matters: This is a price war. Chinese models including DeepSeek are adding further downward pressure that will not ease.

Authentic Brands Group, which controls Elvis Presley's music catalogue and likeness, has filed suit against Anthropic alleging that 274 Presley songs were used to train Claude without permission. The case was filed on Friday.

Why it matters: If rights holders succeed in establishing liability, the costs could be passed upstream.

Jack Clark, co-founder of Anthropic, has revised his forecast for the arrival of recursive self-improving AI from the late 2020s to 2027, citing new benchmark data from P-Zero Research showing that models are improving their ability to design and evaluate their own experiments faster than expected.

The regional angle: Boards that have deferred serious AI strategy on the assumption that transformative capability is still a decade away should revisit that assumption now.

Closing prices, Friday 23 October (US markets). Not investment advice.

Canva has released its own image generation model, now available to all users within the Canva design platform. It promises better quality and customisation than third-party models previously embedded in the tool, at no additional cost for existing subscribers.

THE PLAY: Compare with stock photography and judge whether it reduces your reliance on image libraries.

xAI's Grok platform allows users to generate short-form video content directly from text prompts. Access is available through Grok's web interface.

THE PLAY: Use it to test whether a concept communicates before committing to production spend.

A three-month experiment involving 133 patent attorneys has produced one of the most important findings in AI workforce research this year. Junior lawyers using AI assistance improved immediately and significantly. Their work was faster and more accurate. But when the AI was removed, their performance fell back. Senior attorneys retained the benefit. Junior ones did not.

Senior professionals have enough existing expertise to understand what the AI is doing, catch its errors, and integrate its outputs into genuine knowledge. Junior professionals, still building foundational skills, outsource the thinking that the AI handles rather than learning from it. The tool boosts their output without building their capability.

The same dynamic will appear in finance, consulting, marketing, engineering, and medicine. Organisations that deploy AI as a productivity tool without redesigning how junior talent develops are making a trade: short-term output for long-term capability. That trade may look attractive now. In three to five years, when today's juniors should be filling senior roles, there will be a major talent gap.

Most AI deployment conversations focus on efficiency and headcount. The harder question is what kind of professionals you are producing. If AI handles the difficult parts of a task before a junior employee attempts them, the employee never develops the skill to handle that difficulty independently. They become dependent on AI for a level of performance they can't sustain.

The organisations that will have the most capable senior teams in a decade are the ones that are already asking how their AI tools interact with learning new skills.

When you need to compare options, suppliers, strategies, or candidates, most AI responses default to prose paragraphs that are hard to scan and harder to share. Force a different format by asking for a table and specifying the columns you need.

This works because structuring output as a table forces the model to make explicit trade-offs rather than hedging in prose, and produces something you can paste directly into a board paper or decision memo without reformatting.

Thanks for reading, the wAIve Daily Team. Found this useful? Forward it to someone who needs it. Tell us what you need to know. Reply with your questions.

wAIve Daily is published with Centai Strategy