The AI news blog
AI news, explained by Aleks
Each article takes one real AI news item, links the primary source, and explains what it means for marketers through the lens of my research on user profiling, personality-aware recommendation and explainable AI.
WPP's Q3 2026: The Agency Model Under AI Pressure
WPP reported falling like-for-like net revenue while its CEO pointed to demand for AI-enabled services via WPP Open. What it says about agencies.
Read →Malaysia's AI Governance Bill: Responsibility by Role
Malaysia's draft AI Governance Bill, due by early 2027, puts duties on those who develop, provide or use AI. What role-based rules mean for marketers.
Read →Adobe: AI Traffic Now Converts Better Than Other Visits
Adobe says AI-referred visits to US retailers converted 90% better than other traffic in September. Why AI answers became a channel, and what to do.
Read →Google's AI Mode Ads Arrive in Singapore and Beyond
Google is scaling ads in AI Mode to ten markets, including Singapore, Malaysia and the Philippines. Here is what changes when the ad sits inside the answer.
Read →Agent Skills Goes Vendor-Neutral: Why Marketers Should Care
Anthropic donated the Agent Skills standard to the Agentic AI Foundation. Portable, written-down know-how is exactly what marketing teams need from AI agents.
Read →Meta Q3 2026: AI Lifts Ad Prices, Creators Become a Channel
Meta's Q3 2026: ad revenue up 26%, price per ad up 14%, partnership ads past a $15bn run rate, and a very different Asia-Pacific. What it means.
Read →ChatGPT Ads Become Conversations: Sponsored Agents
OpenAI is testing Sponsored Agents inside ChatGPT and has expanded ChatGPT ads to 89 countries. When an ad can talk back, labelling and honesty become the product.
Read →Anthropic Opens in Singapore: What the MOU Really Signals
Anthropic opened a Singapore office and signed an MOU with MDDI on AI safety and evaluation. Why testing in local languages is the real story.
Read →AI Rules Now Treat Fake Commercial Buzz Like Election Fraud
Anthropic's new usage policy bans deceptive campaigns 'whether political or commercial', and it found fake news made for ad money. A line for marketers.
Read →Workday Buys C1.ai: Your Marketing Agents Need Identities Too
Workday agreed to buy C1.ai, which governs access for people and AI agents. Marketing agents that touch CRM and ad accounts need the same rules.
Read →Character.AI's Chapter 11: Lessons for AI Persona Products
Character.AI filed for Chapter 11 to pursue a sale, with Disney preparing a $20m bid. What brands building AI characters and companions should learn.
Read →Distillation Attacks: What Copied AI Models Mean for You
US agencies named six Chinese AI firms in a distillation advisory, and Anthropic says its defences cut harvested data by 90 to 99%. What this means for buyers of AI.
Read →MAS AI Risk Guidelines: A Useful Template Beyond Banking
MAS issued AI risk management guidelines for financial institutions. Their inventory and third-party rules make a good template for marketing teams.
Read →Claude Haiku 5.5 and the Case for Small Models in Marketing
Anthropic's Claude Haiku 5.5 is built for cheap, fast, high-volume work. Why the smallest model in a family is often the most important one for marketers.
Read →Meta's New AI Ad Tools: Automation Still Needs Explanation
At Advertising Week New York, Meta expanded AI video, its ads assistant, plain-language audiences and an Ads MCP server. What marketers should still own.
Read →AI's Deployment Gap Is a People Problem
Anthropic committed $100m to train 10,000 Frontier Deployed Engineers by 2027 with consultancies and banks. Why marketing needs its own version of this role.
Read →Penske v. Google Dismissed: What It Means for AI Visibility
A US judge dismissed Penske Media's antitrust suit over Google's AI Overviews, finding no 'formal bargain' for traffic. What brands should do about GEO.
Read →Japan's AI Action Plan: Growth Policy Becomes Adoption Policy
PM Sanae Takaichi asked ministers for an AI promotion action plan by year-end, tied to Japan's growth strategy. What adoption-first policy means for marketers.
Read →Gemini 4 Argon: A Million Output Tokens, Now What?
Google DeepMind's Gemini 4 Argon brings a 1M-token output limit and a cyber-first rollout. What long-horizon AI changes for marketing work, and what it doesn't.
Read →Amazon Ads Agent: Retail Media Goes Agentic at unBoxed 2026
At unBoxed 2026 Amazon merged its buying tools into Amazon Ads Agent, with conversational measurement and AI creative. What to keep in human hands.
Read →The White House AI Accord and Independent Evaluators
Tech leaders signed a voluntary White House accord promising outside evaluators, and Anthropic gave METR deep access. Why audits matter for marketing AI.
Read →Why OpenAI Shelved GPT-6.1 Astra, and Why It Matters
OpenAI cancelled the release of GPT-6.1 Astra after testing flagged problems with scope, authorisation and how it reported back. The same tests apply to your agents.
Read →The AI Agent That Wouldn't Take No for an Answer
Australia's prime minister said an OpenAI research agent got around blocks on a Medicare statistics portal. The lesson for marketing agents is about scope.
Read →Claude Opus 5.5: Watermarks, Audits and AI Content Labels
Anthropic's Claude Opus 5.5 ships with watermarking for EU AI Act compliance and a note that the model often suspects it is being tested. Both matter for marketers.
Read →Alibaba's Apsara 2026: Giant Models, Small-Model Reality
At Apsara 2026 Alibaba said it plans a 5 to 10 trillion-parameter model while praising its open 27B model. Why Asia's marketers should watch both ends.
Read →Fields Medallists vs AI Benchmarks: A Lesson for Marketers
Twenty-five Fields Medallists warned that racing AI at unsolved problems confuses a proxy with the goal. Marketing has the same benchmark problem.
Read →Three AI Economic Scenarios and the Future of Marketing Jobs
Anthropic published three scenarios for AI's effect on the US economy to 2030, from modest to extreme. What the knowledge-work findings mean for marketing careers.
Read →Meta Muse: What Happens When Your Customer Sends an Agent
Meta launched Muse, a personal AI agent that can book, shop and negotiate on a user's behalf. Marketing to agents, not only people, is now a practical question.
Read →Korea's Free AI Agents for Every Citizen: A New Front Door
South Korea plans free AI agents for every citizen via KakaoTalk, calls and texts, run by SKT, KT and Kakao. What a public agent layer means for brands.
Read →GPT-6 Astra and the "AGI Era": A Marketer's Reality Check
OpenAI launched GPT-6 Astra and its president said it is not unreasonable to feel we are in the AGI era. What that claim does and does not mean for marketing teams.
Read →