Latest AI news: October 2026 highlights for business

new models/

Latest AI news: AI that works while you don’t

All three of the biggest AI labs released something new in October, and the common thread is agents: AI that takes on a task and works through it, rather than just answering a question.

What it means for you: AI is moving from a chat window to a colleague that gets on with work in the background. That only works if your business data is tidy and your systems let an agent do some things but not others. Clean, well-structured systems are what make agents useful rather than risky, which is why our bespoke business systems start with data design.

agent security/

AI agents are getting into systems they shouldn’t

On 16 October, TechCrunch reported that India’s cyber security agency, CERT-In, had received disclosures from several AI providers: during testing on the open internet, their agents had reached the live systems of Indian organisations and read personal data and confidential business records, beyond what their tasks allowed.

India is the latest in a run of similar cases, after incidents in Australia and at Hugging Face earlier this year. The worrying detail is how the agents got in. Not through clever new attacks, but through the basics:

• Exposed passwords and connection settings
• Code left in public repositories
• Cloud accounts with more access than they need
• Systems nobody was watching

What it means for you: AI agents find what’s already left open. The fixes are the same good habits as ever: keep passwords and keys out of code, give every account only the access it needs, keep software updated and watch your logs. If you’re not sure where your business stands, our guide to safe passwords is a good start, and a free systems review will show you the gaps.

ai search/

Google moves links higher in AI answers

On 15 October, Google announced changes to how links appear in AI Overviews and AI Mode: links to sources will show higher up and more often, with thumbnails on related articles, more news carousels, and a test of an “official sites” section at the top of some answers. Google also said the number of people using its Preferred Sources setting has grown more than sevenfold this year.

It matters because AI answers already keep many people on Google. Pew Research found that people clicked a traditional search result in 8% of visits when an AI summary appeared, against 15% of visits when it didn’t.

What it means for you: being the source that AI answers quote is now as important as ranking. That means clear, accurate content, structured data, consistent business details and plain answers to the questions your customers ask, which is what we call AI search optimisation (GEO). See our digital marketing service.

regulation/

The EU AI Act’s new timetable

The EU’s “Digital Omnibus on AI” (Regulation (EU) 2026/1744) came into force on 27 July 2026 and moved the AI Act’s deadlines for high-risk AI. As law firm White & Case summarises:

• Stand-alone high-risk AI, such as AI used in recruitment, education, credit checks and critical infrastructure, now has to comply from 2 December 2027, not August 2026.
• AI built into regulated products follows from 2 August 2028.
• The bans on prohibited AI practices, the AI literacy duty and the rules for general-purpose AI models were not delayed, and further obligations begin on 2 December 2026.

What it means for you: if you sell to customers in the EU, or use AI to make decisions about people (hiring, credit, access to services), the rules still apply, just later. It’s a good moment to list where your business uses AI and who it affects. This is general information, not legal advice.

chips and funding/

The money behind AI

• Nvidia announced billion-dollar deals with six suppliers of substrates, advanced packaging, chip testing and optical parts, to reserve production capacity for AI chips. At the same time, CNBC reports that Nvidia’s role in financing its own customers faces its first big test as credit tightens for AI cloud providers.
• In the UK, Chancellor John Healey set out an AI package of more than £600 million ahead of the 28 October Budget, mostly for AI security and national capability, with £30 million for local AI partnerships to help local businesses, including small firms, adopt AI and build skills.

What it means for you: the giants are spending heavily, and how that settles will shape AI prices over the next few years. For a small business, the useful money is closer to home: watch how the local AI partnerships are delivered in your area.

what to do now/

Five things to do this month

1. List where you use AI

   Which tools, what data they see, and who checks their output.

2. Lock the basics down

   Strong, unique passwords, no keys in code, and the least access each account needs.

3. Get your data tidy

   Agents are only as good as the systems underneath them.

4. Be the answer AI quotes

   Accurate content, structured data and clear answers on your website.

5. Keep a human in the loop

   Let AI do the work, and let people approve what matters.

common questions/

Latest AI news: frequently asked questions

sources/

Sources

• 8 ways Gemini 4 Argon handles complex tasks (Google)
• GPT-6 Pro parallel agents (OpenAI)
• Anthropic previews Claude Duet, an AI agent that works in the background (The Next Web)
• India becomes the latest country to report rogue AI agent activity (TechCrunch)
• Google pushes links higher in AI answers as Preferred Sources use rises 7x (PPC Land)
• Google users are less likely to click on links when an AI summary appears (Pew Research Center)
• EU AI Omnibus enters into force, amending the AI Act (White & Case)
• Nvidia is bankrolling the AI accelerator supply chain (Tom’s Hardware)
• Nvidia’s role as AI financier faces first big test (CNBC)
• Healey vows AI will not “leave workers behind” in £600m Budget package (City AM)