AI Risk Movers & Shakers - August 2026

19 August 2026 — Industry News, Baseline Updates

By Craig Atkins, Founder, AI Risks

Explore the biggest movers in our August 2026 AI Risks update, as we track changing AI resilience and opportunity scores across industries. From software, professional services and creative industries to education, logistics and skilled trades, discover where accelerating AI agents, automation and robotics are creating the greatest opportunities and the biggest disruption risks.

The biggest theme in our August rebase is that AI opportunity is rising at the same time as AI resilience is falling. That apparent contradiction is increasingly important. More capable reasoning models, multimodal systems and agentic workflows give businesses considerably more scope to automate processes and increase productivity, but those same capabilities lower barriers to entry and threaten businesses selling routine digital or knowledge work. Of the 137 existing macro and industry entries reviewed, we changed the scores of 119. The direction of travel is clearest in technology, professional services, media, education and other sectors where the product of the business can itself be produced or delivered digitally.

Online Tutoring / EdTech is one of the clearest movers. Its opportunity score rises from 9 to 10, while its three-year resilience falls from 5 to 3 and its five-year resilience from 4 to 2. AI tutoring is moving from an interesting supplementary technology towards something capable of providing personalised, always-available instruction at extremely low marginal cost. This creates enormous opportunities for education businesses that successfully incorporate AI, but it poses an equally significant threat to businesses whose proposition is essentially generic one-to-one explanation, practice or course content. Schools and institutions remain more protected by credentials, safeguarding, pastoral care and community, but purely digital tutoring is becoming one of the least resilient sectors in the baseline.

A similar acceleration is visible across software and digital services. Software Development Agencies move from 6/5/4/3 resilience across our 3/5/7/10-year horizons to 4/3/2/1, while their opportunity score increases from 9 to 10. Call Centres / BPO similarly fall from 4/3/2/1 to 3/2/1/1, with opportunity rising to 10. This reflects our growing conviction that the important shift is no longer simply AI helping an employee perform a task faster. Increasingly capable agents can execute longer sequences of work themselves. That changes the economics of businesses built around selling developer hours, support seats, administrative processing or other repeatable digital labour. Complex integration, accountability, security and domain expertise remain defensible, but routine execution is becoming much harder to protect.

Creative and media businesses have also moved down our resilience curve. Marketing & Advertising Agencies fall from 5/4/3/2 to 4/3/2/1, while opportunity rises from 9 to 10. Video Production / Creative Studios make the same resilience move, from 5/4/3/2 to 4/3/2/1, with opportunity increasing from 8 to 10. Publishing and News / Media also become less resilient. Multimodal AI is increasingly capable of producing and manipulating text, imagery, voice and video within the same production workflow. This does not remove the value of creativity, but it dramatically changes the scarcity of production itself. We therefore see value migrating towards distinctive ideas, trusted brands, intellectual property, audience relationships, provenance and exceptional creative direction, and away from charging simply for producing assets.

The other important change is happening outside the office. Warehousing / 3PL falls from 6/6/5/4 resilience to 5/4/3/2, while its opportunity rises from 8 to 9. Courier / Last-Mile Delivery and Freight Forwarding also move lower on resilience and higher on opportunity. We remain cautious about assuming rapid, universal deployment of humanoid robots or fully autonomous transport, but the medium-term robotics outlook has strengthened enough to affect our five, seven and ten-year assumptions. Computer vision, autonomous navigation and increasingly general robotic systems mean that physical work can no longer automatically be treated as permanently insulated from AI. The distinction is increasingly between structured, repeatable physical environments, such as warehouses, and messy, variable environments such as homes, construction sites and emergency situations.

That distinction explains why many skilled trades, healthcare and hands-on services remain among the more resilient parts of the baseline. Electricians and plumbers remain at 9/9 over the three and five-year horizons, for example, even though we now allow resilience to decline further over the longer term. AI can transform quoting, diagnostics, scheduling, compliance and customer service without eliminating the need for somebody to physically install, repair or certify equipment. Healthcare follows a similar pattern. AI opportunity is increasing as clinical support and administration improve, but regulation, liability, safeguarding, physical procedures and human care remain powerful constraints on substitution. Our horizon scanning therefore does not suggest that "AI replaces everything". Instead, the emerging divide is between work that AI can generate, decide and execute digitally, work that increasingly capable machines can perform in controlled physical environments, and work that still depends on human presence, dexterity, trust, accountability or unpredictable real-world problem solving.

Taken together, the August update represents a shift in how we think about the next decade. The central question is becoming less "Can AI help this industry?", because the answer for almost every industry is now yes, and more "What remains scarce when intelligent digital labour becomes cheap?" Businesses built around scarce physical assets, regulated accountability, trusted relationships, proprietary data, distinctive intellectual property and difficult real-world execution generally retain stronger resilience. Businesses whose value is primarily the production, transformation or movement of information face a much steeper curve. At the same time, some of those vulnerable sectors have the highest opportunity scores in the entire baseline. The winners may therefore not be the industries least affected by AI, but the businesses that recognise the disruption early enough to become the ones applying it.