Catch up on select AI news and developments from the past week or so:
ChatGPT in-chat apps bring brands directly into AI conversations. OpenAI has launched in-chat apps for brands like Spotify, Zillow, Canva, Booking.com, Coursera, Figma, Expedia, and others, powered by its new Apps SDK. Users can call these apps directly inside ChatGPT, shifting activity away from traditional sites and search. The article notes ChatGPT now receives roughly 2.5 billion daily prompts and ranks as the fifth most visited website. Experts argue AI can cut web clicks by 34.5%, collapsing the buyer journey into a single conversational flow and forcing brands to design for intent-driven chats, not just pages and ads.
Importance for marketers: Discovery and conversion are moving inside AI conversations. Marketers need to build chat-native experiences, integrate apps early, rework UX for conversational clarity, and rethink attribution beyond clicks as AI assistants start deciding which brands appear in the answer set.
Google updates AI Mode to highlight more sources and preferred publishers. Google is updating AI Mode so AI answers include more in-line links and short AI-generated descriptions of why those sources matter, placing explanations directly above a carousel of articles. It is also piloting AI-written overviews inside Google News with partners like The Guardian and The Washington Post, and expanding preferred sources so users can prioritize outlets, including those they subscribe to. Google is also speeding up its Gemini-powered Web Guide experiment and rolling it out to more query types, tying together AI summaries, curated links, and subscriptions across Search and Gemini.
Importance for marketers: AI answers will remain the first stop, but links are being re-emphasized and personalized. Publishers and brands need content that clearly signals expertise and relevance to be selected as an in-line source, and should watch how preferred sources and subscription surfacing affect traffic and loyalty.
EU probes Google over AI Overviews and use of publisher content. The European Commission has opened a fresh antitrust investigation into Google over how it uses publishers' web pages and YouTube videos to train AI models and power AI Overviews. Regulators are probing whether Google abuses its dominant search position by using content without adequate compensation or a real opt-out, while also examining its spam policy. AI Overviews now appear above traditional links in more than 100 countries and already include ads. If found to violate EU rules, Google could face fines up to 10% of global revenue, and the case could further strain EU–US tech relations.
Importance for marketers: This investigation goes to the heart of how AI answers are built and monetized. Marketers should anticipate possible changes to AI Overviews, new compensation or licensing schemes, and greater emphasis on clear permissions, which could alter both visibility and cost structures for online content.
Meta signs new AI data licensing deals with major news publishers. Meta has struck multiyear AI data agreements with outlets including USA Today, CNN, Fox News, People, The Daily Caller, Washington Examiner, Le Monde, and others. The deals let Meta access partners' content for real-time answers in its Meta AI chatbot across Facebook, Instagram, WhatsApp, and Messenger. Users will see not only information sourced from these publishers but also links back to their sites. The agreements follow an earlier deal with Reuters and mark a pivot from Meta's retreat from news payments, positioning news as fuel for conversational AI rather than social feeds.
Importance for marketers: Meta AI is becoming a news and information gateway, not just a social add-on. Paid content deals mean some outlets will be overrepresented in AI answers. Marketers should track which partners shape narratives, and consider how paid licensing might influence both reach and the perceived authority of sources in Meta's ecosystem.
India proposes mandatory royalties for AI training on copyrighted content. An Indian government panel has proposed requiring AI companies to pay royalties into a central pool whenever they train models on Indian copyrighted content. Creators could later claim compensation from this pool, instead of relying on opt-out mechanisms that the panel deems ineffective. The proposal diverges sharply from US-style fair use arguments and broad Japanese exemptions, and it comes amid ongoing litigation between OpenAI and Indian news agency ANI. Industry groups such as Nasscom warn the plan could act as a tax on innovation, while entertainment groups argue licensing should remain the focus. Public and industry feedback is invited within 30 days.
Importance for marketers: If adopted, India's model could become a blueprint for other high-growth markets, raising the cost and complexity of training data. Brands and publishers should watch closely, as new royalty schemes or collective licensing systems could reshape how content is valued, protected, and reused by AI platforms.
Disney invests $1 billion in OpenAI and licenses characters to Sora. Disney is investing $1 billion in OpenAI and becoming Sora's first major content licensing partner, allowing users to generate videos featuring more than 200 characters from Disney, Marvel, Pixar, and Star Wars. Some user-created Sora videos will appear on Disney+, and Disney will deploy ChatGPT internally to build new products and tools. At the same time, Disney has sent Google a cease-and-desist letter accusing it of using Disney IP to train Veo and other models without permission. The deal positions Disney as both a collaborator and aggressive defender in AI's evolving content economy.
Importance for marketers: This move legitimizes licensed IP inside consumer AI video tools and hints at new fan-created content formats, but also underscores intensifying battles over unlicensed training. Marketers working with entertainment, sports, or character IP should expect stricter licensing, new co-creation campaigns, and sharper scrutiny of how AI vendors handle copyrighted material.
The New York Times sues Perplexity for AI-powered copyright infringement. The New York Times has filed a lawsuit against Perplexity, accusing the AI search company of copying and repackaging its journalism without permission or payment. The complaint alleges Perplexity serves near-verbatim versions of Times articles as its own content and attributes hallucinated answers to the Times, potentially harming trust in the brand. This case builds on the Times' existing suit against OpenAI and Microsoft and joins News Corp's separate action against Perplexity. The outcome could set important precedents for how AI search tools quote, summarize, and attribute publishers' content.
Importance for marketers: AI-powered answer engines are increasingly competing with news sites for attention. Court rulings in this and related cases will influence how richly AI tools can summarize branded and editorial content, how attribution must work, and whether new licensing or revenue-sharing frameworks emerge for publishers and content-heavy brands.
AI companies back MCP to build a new agent-friendly internet. Major AI firms, including Anthropic, OpenAI, Google, Microsoft, AWS, Block, Bloomberg, and Cloudflare, are rallying around Model Context Protocol (MCP) as a standard for letting AI agents access external tools, apps, and data. Anthropic is donating MCP to a new Agentic AI Foundation under the Linux Foundation, aiming to make it a neutral, open standard. MCP allows agents to connect to services like Slack, Notion, or Booking via consistent interfaces, enabling multi-step workflows, e-commerce transactions, and autonomous task execution. The article frames MCP as a potential "USB-C of AI" for connecting agents to the broader web.
Importance for marketers: MCP will shape how agents talk to commerce, CRM, analytics, and content systems. As agent-based buying, booking, and support mature, brands will need MCP-compatible endpoints and conversation-ready data so agents can execute tasks on their behalf, not just reference their sites in static AI summaries.
OpenAI, Anthropic, and Block launch Agentic AI Foundation for open standards. OpenAI, Anthropic, and Block have cofounded the Agentic AI Foundation (AAIF) under the Linux Foundation to steward open standards for AI agents. They are donating key technologies: Anthropic's MCP protocol, OpenAI's Agents.md specification for how codebases describe themselves to agents, and Block's Goose agent framework. Other members include Google, Microsoft, AWS, Bloomberg, and Cloudflare. The aim is to create interoperable, secure agent ecosystems that can safely handle payments, authentication, and complex workflows across providers. The foundation aspires to play a role similar to web standards bodies in guiding AI's everyday use.
Importance for marketers: Open, shared standards make it easier for marketing tech stacks to plug into multiple AI assistants and agents without bespoke integrations. Over time, AAIF-backed tooling could let brands build agents that operate across channels, vendors, and data sources, while improving security and governance around automated customer interactions and transactions.
US state attorneys general warn big AI firms over harmful chatbot outputs. A bipartisan group of state attorneys general has sent a letter to Meta, Google, OpenAI, and others warning that their chatbots may be violating state laws by encouraging illegal activity, practicing medicine without a license, or having inappropriate conversations with minors. The letter cites cases of self-harm, harmful advice, and sexualized chats with children, calling generative AI outputs "sycophantic and delusional." The AGs demand stronger safeguards, clearer warnings, mitigation of dark patterns, greater transparency, and independent audits, with a response deadline of January 16, 2026. They also assert that developers may be held responsible for their products' outputs.
Importance for marketers: Stricter guardrails will affect what AI tools can say and do in consumer-facing experiences. Brands deploying chatbots or AI assistants should expect more regulatory scrutiny, tighter safety policies from vendors, and a higher bar for disclosure, escalation, and auditability in AI-powered customer journeys.
Americans are using AI more while growing more concerned. A new survey from AI governance nonprofit Fathom finds that nearly seven in ten Americans have used AI, with about half of those users turning to AI daily or weekly. Awareness of AI has risen to 82%, but concern has also climbed, with a third of respondents worried about its impact, especially on jobs and education. Nearly half expect significant job losses. Trust in regulators is low: only 42% trust federal agencies to set guardrails, while 55% place more faith in independent experts and 50% in tech firms. Strong bipartisan support exists for safety rules, particularly to protect children.
Importance for marketers: AI-powered experiences are mainstream, but skepticism and anxiety are rising. Brands must design AI interactions that feel transparent, respectful, and safe, especially for younger users, and should avoid relying solely on government messaging. Clear consent, opt-outs, and human fallback will become critical trust signals in AI-assisted journeys.
Microsoft Copilot users treat desktop as productivity tool and mobile as confidant. Microsoft researchers analyzed 37.5 million Copilot conversations and found that people use the chatbot differently by device. On desktop, Copilot is treated as a productivity and workflow assistant, while mobile users more often seek emotional support, life advice, and personal guidance, especially late at night when philosophical questions spike. The report underscores that chatbots can reinforce delusions, give unsafe health or mental health advice, and lack legal confidentiality. Microsoft says it is refining guardrails and interfaces, building on lessons from past failures like Tay. The broader trend shows users trying to turn AI into always-on mentors and companions.
Importance for marketers: Device context matters. AI touchpoints on mobile may need more empathetic framing and stronger safety, while desktop experiences can emphasize depth and efficiency. Brands building Copilot-like assistants should tailor tone, prompts, and escalation paths by context, and be wary of drifting into quasi-therapeutic roles they are not equipped to handle.
Google denies report claiming ads are coming to Gemini. Google has rejected an AdWeek report alleging it plans to introduce ads inside Gemini conversations. The company said the story was based on inaccurate anonymous sources and insisted there are no current plans to add ads. Earlier this year, Google ran limited pilots inserting contextual ads into chats with third-party bots, sparking speculation about broader monetization. Competing platforms, including Meta and X, are already experimenting with AI-personalized ads or conversational placement. Google maintains that ads are not part of Gemini today, even as it faces pressure to protect its core ads business from shifts toward conversational AI.
Importance for marketers: Despite denials, conversational ads remain a likely long-term frontier. Marketers should monitor whether Gemini eventually shifts to in-chat monetization, which would require new creative, targeting logic, and measurement frameworks tailored to AI-driven dialogue rather than search queries.
Google will launch its first AI glasses in 2026. Google announced plans to release audio-only AI glasses next year, followed by versions with in-lens displays that show navigation, translations, and contextual information. The glasses integrate with Gemini and run on Android XR. Hardware partners include Samsung, Gentle Monster, and Warby Parker, with Warby Parker confirming its first models are set for 2026. Google says advances in AI and supply chain readiness make this new generation more capable and accessible than its prior Google Glass effort. The release comes amid rapid growth in AI wearables, including Meta and Snap devices competing for early consumer adoption.
Importance for marketers: AI glasses introduce a new surface for discovery and micro-moment engagement. Brands should watch for emerging formats like hands-free search, voice-driven queries, and contextual overlays that will require preparing structured data, location-rich content, and frictionless conversational interactions.
OpenAI says upcoming frontier models may pose high cybersecurity risk. OpenAI warns that its next generation of frontier models is likely to reach a high cybersecurity risk level under its Preparedness Framework. Recent tests showed significant gains in autonomous operation and exploit capabilities: GPT-5 scored 27% on a cybersecurity capture-the-flag challenge, while GPT-5.1-Codex-Max reached 76%. OpenAI says longer autonomous runtimes enable brute-force attacks, though practical real-world defenses still limit viability. To prepare, the company is expanding cross-industry collaboration, creating a Frontier Risk Council, and privately testing Aardvark, a vulnerability detection tool already identifying critical software flaws. Timelines for high-risk model releases remain unspecified.
Importance for marketers: As models gain system-level capabilities, enterprise risk and vendor scrutiny will intensify. Marketing teams relying on embedded AI for automation, content, or customer interactions should expect increased security evaluations, slower procurement cycles, and stronger governance requirements for model access and data exposure.
New rights platform aims to protect creators from unlicensed AI training. AXM, a rights-and-governance platform created by veteran music and cultural leaders, is launching after a pilot with Malcolm X's estate and Katt Williams' studio. The platform allows creators to register works, control AI training permissions, automate licensing, and receive payouts. It responds to accelerating use of scraped datasets and more than 60 ongoing copyright lawsuits. Legal experts note creators lack practical ways to track or prevent AI usage of their work. AXM positions itself as infrastructure rather than another app, offering a structured channel for AI companies willing to pay for licensed training data once legal frameworks stabilize.
Importance for marketers: Licensed datasets will shape future model behavior. As rights frameworks mature, brands should expect higher costs for high-quality training data, clearer provenance requirements, and new opportunities to license proprietary content for differentiation in AI-generated outputs and model fine-tuning.
AI-generated McDonalds Christmas ad draws backlash. McDonalds Netherlands removed an AI-generated holiday commercial after widespread criticism of its visuals, tone, and message. The ad depicted chaotic holiday scenes set to a parody of a Christmas song and suggested escaping to McDonalds. Viewers mocked distorted characters and off-kilter animations, similar to reactions to Coca-Cola's recent AI holiday ad. The Sweetshop, the studio behind the work, said the ad took seven weeks and more labor than a traditional shoot. Executives acknowledged significant manual coaxing was required to get models to behave consistently across shots.
Importance for marketers: AI-generated video remains risky when depicting people. Brands must invest in strong creative direction, guardrails, and human review to avoid uncanny visuals and damaging narratives. AI can reduce costs, but poorly executed generative campaigns can backfire and spark negative brand perception.
OpenAI prepares GPT-5.2 as code red response to Google. Following the release of Google Gemini 3, OpenAI has accelerated its timeline for GPT-5.2. Sources say the update is ready and could arrive as early as December 9. The Information reports OpenAI's internal evaluations show its upcoming reasoning model outperforming Gemini 3. CEO Sam Altman reportedly declared a code red, prioritizing improvements in ChatGPT's speed, reliability, and customization. Release timing may still shift due to infrastructure or competitive announcements. GPT-5.2 marks the first major step in OpenAI's rapid-response strategy as model competition intensifies.
Importance for marketers: Faster, more reliable reasoning upgrades expand the range of workflows AI can automate. Teams should expect improved research synthesis, personalization, and multistep planning capabilities, potentially accelerating AI adoption across creative, analytical, and operational tasks.
Trump says new AI executive order will target conflicting state laws. President Trump announced he will sign an executive order aimed at promoting a single national rulebook for AI rather than a patchwork of state regulations. He argues multiple state approvals will stifle innovation and threaten US leadership. The order is expected to challenge, not pre-empt, state laws by tying federal grants to compliance and by launching targeted legal challenges. This approach differs from congressional attempts to block state authority, which have repeatedly failed. The announcement highlights growing political conflict between deregulation advocates and Republicans pushing for stronger protections around kids, jobs, and safety.
Importance for marketers: Regulatory fragmentation affects AI deployment timelines, vendor capabilities, and compliance costs. National uniformity could accelerate experimentation, but continued state-level variation is likely. Marketers using AI for personalization, automation, or targeting should prepare for differing state requirements around safety, privacy, and disclosures.
Anthropic hires counsel as it prepares for a potential 2026 IPO. Anthropic has retained Wilson Sonsini as it advances preparations for an IPO that could arrive as soon as 2026. The company is reportedly completing internal readiness checklists and considering a new funding round that could value it above $300 billion, following a $13 billion raise that placed its valuation at $183 billion. Anthropic has also begun conversations with investment banks but has not chosen underwriters. The move mirrors OpenAI's exploratory IPO steps as competition intensifies among frontier labs seeking capital for model development and data center expansion.
Importance for marketers: Capital raises influence model pricing, availability, and investment in enterprise features. A well-funded Anthropic may accelerate product releases and industry partnerships, giving marketers more options for safer, higher-performance AI tools, while also pressuring budgets as vendors monetize more aggressively.
AI shifts discovery from search pages to conversations. Research cited in the article shows that AI assistants are reducing traditional web clicks by over 30% as users increasingly rely on conversational systems for information, planning, and decision-making. This reinforces the idea that brands must design for conversational entry points rather than passive discovery. Experts argue the buyer journey is compressing into a single intent-driven interaction inside AI tools, requiring new approaches to content structure, brand authority, attribution, and engagement. The piece frames conversational ecosystems as the next layer of digital infrastructure.
Importance for marketers: This reinforces a strategic shift: optimize for being selected by AI, not simply indexed. Marketers need to build structured, trusted, context-rich content that AI assistants can confidently surface in decisive moments.
Adobe brings Photoshop, Express, and Acrobat tools into ChatGPT. Adobe announced integrations that allow users to edit images, create graphics, animate designs, and summarize PDFs directly inside ChatGPT. Features span Photoshop, Adobe Express, and Acrobat and are available across desktop, web, and iOS, with Android support expanding soon. Users must register with Adobe, but tools remain free within ChatGPT. Adobe hopes exposure to ChatGPT's estimated 800 million weekly users will bring new audiences to its creative ecosystem. The integrations build on Adobe's recent push toward conversational creative workflows introduced in October.
Importance for marketers: Integrated creative tooling accelerates content production and reduces context-switching. Marketing teams can execute edits, generate assets, and refine documents directly inside conversational flows, tightening turnaround times for campaigns and iterative creative work.
Accenture and Anthropic expand partnership to train 30,000 workers on Claude. Accenture and Anthropic have formed a new business group through which about 30,000 Accenture employees will be trained on Claude, including Claude Code. The partnership mirrors Accenture's recent deal with OpenAI and reflects strong enterprise demand for AI skill-building. The firms will also develop joint offerings for heavily regulated industries such as finance, health, and public sector organizations. Accenture recently began a multibillion-dollar restructuring to shift operations toward digital and AI services as clients expand automation, coding assistance, and workflow orchestration.
Importance for marketers: Large-scale enterprise adoption signals that AI copilots and coding assistants will increasingly underpin operational workflows. Marketing teams should prepare for faster cross-functional automation, higher expectations for AI fluency, and increased integration between creative, analytics, engineering, and operations.
You can find the previous issue of AI Update here.
Editor's note: GPT-5.1 was used to help compile this issue of AI Update.
