Catch up on select AI news and developments from the past week or so:
OpenAI launches Sora 2 with audio and likeness features. OpenAI unveiled Sora 2, a video generation model that supports synchronized audio and likeness integration. Users can create videos with dialogue and sound effects, and upload identity-verified recordings to generate "cameos" of themselves or friends. Alongside the model, OpenAI released a social video app where users remix clips in TikTok-like feeds. The launch comes amid intensifying competition with Meta, Google, xAI, and Anthropic. Importance for marketers: Sora 2 expands branded storytelling opportunities with immersive, customizable short-form video formats. A new social channel centered on AI-generated video, reshaping influencer marketing and brand engagement.
ChatGPT adds instant checkout for in-chat shopping. OpenAI introduced an Instant Checkout feature that lets US ChatGPT users purchase products directly in chat. Initially supporting Etsy sellers, the tool will soon extend to Shopify merchants including Glossier, SKIMS, and Spanx. Transactions run on the open-sourced Agentic Commerce Protocol, allowing secure order transmission while merchants handle fulfillment and support. Merchants pay a small fee per sale, but product pricing and ranking remain unaffected. The feature positions ChatGPT as both a discovery and transaction hub, deepening user reliance on AI-driven commerce. Importance for marketers: This signals a major shift toward conversational commerce, requiring brands to optimize content for AI-driven shopping flows.
Microsoft adds Agent Mode to Excel and Word in Copilot. Microsoft introduced Agent Mode in Excel and Word plus Office Agent in Copilot chat, enabling multistep automation of spreadsheets, reports, and presentations. Built on OpenAI's reasoning models, the tools democratize advanced Excel modeling and streamline Word authoring through iterative dialogue. Office Agent generates polished PowerPoint decks from chat prompts, integrating deep web research. Early benchmarks show significant accuracy gains in complex tasks. Agent Mode is rolling out in the Frontier program, with desktop availability coming soon. Importance for marketers: Marketers can now scale reporting, analysis, and content creation, reducing manual effort in data-driven campaigns and presentations.
California enacts first US AI safety transparency law. Governor Gavin Newsom signed SB 53, the Transparency in Frontier AI Act, requiring AI companies to report safety incidents and publish documents on best-practice safeguards. The law applies to California-based leaders like OpenAI and Anthropic, impacting 32 of the world's top 50 AI firms. Noncompliance carries civil penalties, and whistleblower protections are included. The bill marks the first AI law explicitly focused on frontier model safety in the US, with global ripple effects expected. Federal lawmakers are also considering similar mandates. Importance for marketers: Stricter AI transparency rules will influence vendor practices, compliance messaging, and risk management in martech partnerships.
Anthropic launches Claude Sonnet 4.5 with stronger coding and safety. Anthropic released Claude Sonnet 4.5, described as its smartest and safest model yet. The update enhances coding, cybersecurity, finance, and research capabilities, while offering extended 30-hour autonomous task execution. It surpasses prior models in accuracy and reduces risks such as deception and sycophancy. Benchmarks like SWE-bench Verified rank it as a top coding model. The launch continues Anthropic's rapid innovation cycle, outpacing rivals in specialized enterprise use cases. Importance for marketers: Claude Sonnet 4.5 enables faster, safer automation of technical workflows, offering marketers improved tools for secure data handling and campaign execution.
Anthropic to triple global workforce as Claude demand surges. Anthropic plans to triple its international staff and expand its applied AI team fivefold to meet soaring global demand. Nearly 80% of Claude usage now occurs outside the US, with heavy adoption in Asia-Pacific and Europe. Revenue run-rate hit $5 billion in August, up from $1 billion earlier this year, with enterprise customers growing from under 1,000 to over 300,000 in two years. New offices are opening in Tokyo, Dublin, London, and Zurich. Importance for marketers: Global expansion ensures broader regional support and localization for marketers using Claude in international campaigns.
Opera debuts Neon AI browser for agentic web use. Opera launched Neon, an AI-powered browser with features like form-filling, cross-site data comparison, and in-browser coding. The Neon Do feature lets the browser act on behalf of users without external cloud processing. Privacy-focused, on-device operations target European regulators' concerns. Features like "Tasks" and "Cards" streamline workflows with reusable prompts and AI workspaces. Neon is subscription-based, launching in early access. Importance for marketers: Neon reflects the growing trend toward agentic browsing, requiring marketers to anticipate AI-driven interactions that bypass traditional search and navigation.
Amazon unveils AI-powered Kindle, Echo, and Ring devices. Amazon launched new AI-driven versions of Kindle Scribe, Echo speakers, Ring cameras, and Fire TVs, all powered by Alexa+. The upgrades add personalized experiences, facial recognition, and conversational search. The Kindle lineup now includes lighter models with AI-powered search and a new color option. New Echo devices feature deeper personalization and insights, while Ring cameras integrate family recognition. Panos Panay, leading Amazon devices, emphasized seamless AI integration into everyday life. Importance for marketers: Amazon's deeper AI integration creates more personalized touchpoints for retail and advertising, reinforcing Alexa+ as a key ecosystem for consumer engagement.
Samsung tests AI-powered notification summaries in One UI 8.5. A leaked One UI 8.5 build confirms Samsung is adding AI notification summaries, though the feature isn't fully functional yet. Built on Google's Gemini Nano model, it summarizes long notifications without sending data to external servers. Users will be able to exclude apps from summaries once the feature is complete. This aligns with Samsung's broader Galaxy AI push across devices. Importance for marketers: Notification summaries may change how users consume mobile alerts, requiring marketers to craft concise, high-value messages optimized for AI filtering.
OpenAI launches parental controls after teen suicide case. OpenAI introduced parental controls for ChatGPT following a lawsuit linked to a California teen's suicide. Parents and teens can link accounts for stronger safeguards, including blocking voice mode, restricting sensitive content, and setting quiet hours. OpenAI is also building an age prediction system to auto-apply teen settings. Regulators are scrutinizing AI firms' responsibilities for child safety, echoing Meta's recent safeguards. Importance for marketers: Stricter parental controls highlight growing regulatory oversight, shaping how marketers target and engage younger demographics responsibly.
Nothing launches Playground for AI mini-app creation. Smartphone maker Nothing introduced Playground, an AI tool allowing users to create mini-apps (widgets) via text prompts. Apps can track flights, summarize meetings, or serve as virtual pets. More technical users can edit code for customization. Nothing aims to pioneer "vibe coding" for mobile, positioning AI as a new OS layer. Security and ease-of-use remain priorities, with no paid tier yet. Importance for marketers: Playground could evolve into a micro-app marketplace, giving brands new interactive formats embedded in consumers' daily device use.
Stravito introduces AI Personas for market research insights. Knowledge management firm Stravito launched AI Personas, enabling teams to query consumer research through conversational profiles representing demographics or segments. Marketers can test campaign ideas against dynamic personas, extracting localized insights quickly. Companies like Lavazza report reduced validation times and stronger consumer-centric strategies. Governance controls ensure data traceability and security. Importance for marketers: This offers faster, deeper audience insight extraction, allowing for more consumer-driven creative development and campaign optimization.
China's DeepSeek releases intermediate AI model. DeepSeek launched V3.2-Exp, a transitional large language model featuring Sparse Attention for efficiency and lower costs. It reduces API pricing by over 50% while improving long-sequence processing. The model is framed as a step toward its next-generation architecture, following disruptive launches earlier this year. Analysts expect competitive pressure on Alibaba's Qwen and Western rivals. Importance for marketers: Cheaper, efficient models like DeepSeek's may lower campaign automation costs and increase options for localized AI adoption in Asia.
Google blocks AI search results on Trump dementia queries. Google's AI Overviews decline to summarize queries about Donald Trump and dementia, while providing responses for similar queries on Biden and Obama. Instead, searches yield only standard web results. The move highlights the sensitivity of political health topics and the risks of AI-generated misinformation. A Google spokesperson said AI Overviews do not appear for every query. Importance for marketers: This underscores evolving boundaries in AI-generated search, reminding brands to anticipate uneven visibility and moderation policies across sensitive topics.
You can find the previous issue of AI Update here.
Editor's note: GPT-5 was used to help compile this issue of AI Update.
