AI Daily Brief: 26 October 2026

26 October 2026

Today's AI news puts business value and control in focus. An economist questions whether spending is reaching company profits, while fresh disputes over copyright, search traffic and cross-border model access show why adoption needs more than a technology budget.

Apollo economist sees no AI payoff in non-tech company margins

Apollo Global Management chief economist Torsten Slok says profit margins and revenue per worker appear flat for the S&P 493, the S&P 500 excluding seven large technology companies. He argues there is no visible financial payoff from AI for corporate America outside technology, despite heavy investment.

This is a broad market observation, not proof that every individual AI project fails. UK firms should judge their own deployments against a baseline for cost, quality, turnaround time and revenue rather than treating tool adoption as the result.

Our take: Measure the whole workflow, including review and rework. A faster draft is not a saving if a skilled employee spends the recovered time checking errors.

Elvis Presley rights owner sues Anthropic over 274 songs

Authentic Brands Group, which controls Elvis Presley's music and likeness rights, has accused Anthropic of using 274 Presley songs to train AI models without permission. The complaint seeks up to $150,000 per work, about $41 million in total. These are allegations, not a court finding.

For businesses buying generative tools, the case underlines the difference between a vendor offering a model and a buyer having clear rights to use its outputs in commercial campaigns. Review licensing terms and keep a human check on recognisable lyrics, voices and likenesses.

Our take: A tool's availability is not a copyright clearance. Marketing teams need a documented route for approving AI-generated creative work, especially when it resembles a known artist.

China reportedly lets DeepSeek and MiniMax continue AI exports

After adding 23 technologies related to large models, agents, computing clusters and information processing to an export-control list, China has reportedly authorised DeepSeek and MiniMax to continue exporting. The Register reports that technology issued under open-source licences is exempt from the restrictions.

UK organisations evaluating Chinese AI suppliers should distinguish model availability today from long-term contractual access. Export permissions, hosting location and support obligations belong in procurement due diligence.

Our take: Do not make an AI workflow dependent on one provider's uninterrupted cross-border access. Test whether you can move prompts, evaluations and data to an alternative model.

AI model price cuts sharpen the cost-per-task question

Researcher Max Woolf has traced recent OpenAI and Anthropic API pricing changes, including an 80% cut to GPT-5.6 Luna input and output prices and a 75% cut to Fable 5.1 cache-read pricing. His analysis argues that cheaper open-weight alternatives are putting pressure on proprietary model prices.

Headline token rates rarely show the full bill for an agent that makes repeated calls, reads long context and needs retries. UK buyers should compare cost per successful task across models, including staff review, before locking in an architecture.

Our take: Re-benchmark production workflows when prices change. A cheaper token can be valuable, but only if completion quality and total calls remain acceptable.

AI search is changing the traffic bargain for publishers

A new Verge analysis traces how Google's AI Overviews and AI Mode have changed the relationship between search engines and publishers. It cites a reported fall in the share of New York Times visits from search, from 44% to 35.5% over three years, as one sign of a wider traffic shift.

The figures do not isolate AI as the sole cause. Still, UK publishers and businesses relying on informational search traffic should monitor impressions, clicks and conversions separately, then build direct audiences through email and useful owned content.

Our take: Being cited in an AI answer and earning a website visit are different outcomes. Track both where possible and keep channels that connect you directly with customers.

Grok publishes seven minutes of its AI-made Odyssey

Grok has released seven minutes of an AI-generated Odyssey project, now described as the first episode of a series rather than the full-length film previously promised. The release shows a public example of AI video production, but not a completed feature.

For UK creative teams, a short demonstration is worth testing against real production requirements: consistent characters, editability, rights clearance and audience response. It should not be treated as evidence that a full production process has been replaced.

Our take: Judge generative video on the finished deliverable, not on a compelling clip. The gap between a preview and repeatable commercial production remains material.

Frequently Asked Questions

How often is the AI Daily Brief published?

Every morning at 7:30am UK time, covering the previous 24 hours of AI news from over 30 sources.

How are stories selected?

UK-relevant stories are prioritised first, then by business impact and practical implications for UK organisations adopting AI.

Why should business leaders follow AI news?

AI is moving faster than any technology in history. Staying informed is essential for making smart decisions about AI investment, adoption, and governance.