The Captain's Log
Jeremy Rosenberg
Generative AI, Marketing, Brand & Communications Consultant, Trainer & Coach
AI Newsletter 13 July 2025
🦜 Aye, Aye, AI
Your weekly GenAI briefing for people who think, create or lead.
Edition: 14 July 2025
Welcome to your weekly intelligence download. This week, Adobe and HeyGen unveil AI tools that behave like full creative teams, Mary Meeker - yes The Mary Meeker - drops her first AI-only report, and a TikToker teaches GPT‑4o to flirt in her DMs — then lets it take over. Of course she did.
In this week’s briefing:
AI and Content Creation
AI Tools, Models and Platforms
AI, Marketing and Sales
AI Adoption
Policy, Power and Ethics
And Finally…
1. AI and Content Creation
Welcome to your AI production crew.
• HeyGen’s new “Creative Operating System” is here: Think of it as an AI agent for full video production: it takes your idea, builds a script, selects visuals, casts AI actors and delivers finished ads or reels or “UGC”. This is no longer coming — it’s here and available now. See HeyGen’s launch video →
• Moonvalley opens Marey AI video model to public: This LA-based startup is courting creators by offering directorial controls: change camera angles, edit motion, and keep it copyright-safe. Try Marey and see the demo →
Google adds image-to-video to Veo 3 — unless you're based in the UK, Switzerland or the EU. Google has started to roll out a new image-to-video feature in Veo 3 via the Gemini app and Flow tool. This allows Pro/Ultra subscribers in 150+ countries (lucky you) to upload a still image and generate an 8-second video with native audio. While, the feature remains unavailable in the UK, Switzerland and the EU, users there can still generate videos from text prompts. All outputs include visible watermarks and SynthID tagging for transparency. Read TechCrunch’s full report → Read Google’s announcement →
2. AI Tools, Models and Platforms
• OpenAI’s browser is reportedly in testing — and Operator may power it: Early users have spotted a new browser option inside ChatGPT. It may integrate Operator, the company’s in-house web agent, to turn browsing into something more like delegating. Read Tom’s Guide report →
• Perplexity officially launches Comet AI browser: The “anti-Google” search startup now has its own browser, built from the ground up to support AI-first tasks and everyday workflows. Read The Decoder coverage →
• GPT‑5 is believed to be coming — and soon: OpenAI has confirmed that GPT‑5, described as a unified “omnimodel” capable of handling text, audio, image — and possibly video — is expected this summer (mid‑2025). It promises smarter reasoning, tool use, and deeper multimodal integration. Read the Techzine EU report →
• Grok 4 launches with a $300/month “SuperGroks” tier: Elon Musk’s new AI chatbot also offers a heavy-duty version aimed at power users with enterprise tasks deep pockets — but it launches under a cloud of controversy after earlier antisemitic and neo-nazi outputs and the departure of X CEO (now X ex-CEO), Linda Yaccarino, after just two years. Read The Verge coverage →
3. AI and Marketing and Sales
AI is already behind the biggest campaigns you’ve seen.
• Unilever ’s Dove AI campaign hits 3.5 billion views: Dove’s long-running campaign promoting real beauty now uses generative AI to generate and scale content globally to equip its army of influencers and deploy "thousands of assets a week across our brands, compared to single digits over months". Read the WSJ feature →
4. AI Adoption
AI is no longer an experiment — it’s infrastructure.
• Mary Meeker’s 2025 AI report lands with force: At 340 slides, it’s not light reading. But Meeker’s (yes, that Mary Meeker) first AI-only trend report makes a powerful case for agentic systems, computational labour units, and the evolution of interface design — the very shifts now showing up in tools like Adobe’s brand agents and HeyGen’s video OS. Read the full report →
5. Policy, Power and Ethics
The world is not ready for election-year deepfakes.
• The United Nations’s International Telecommunication Union has sounded the alarm: a surge of sophisticated deepfakes is expected to flood global media throughout the 2025/26 election cycle — yet most governments aren’t prepared. A report released at the ITU’s AI for Good Summit in Geneva this month, warns that AI-generated videos, audio and images could impact public trust and election integrity, and urges adoption of watermarking, provenance standards and platform cooperation to counter the threat. Read Reuters coverage →
6. And Finally…
Because nothing says 2025 quite like AI impersonating politicians...
AI impersonator targets foreign leaders — using Rubio’s voice: According to a new AP News report, someone used synthetic audio and chat to impersonate U.S. Secretary of State Marco Rubio — sending voice messages, texts and Signal invites to at least three foreign ministers, a U.S. senator and a governor. The State Department has issued a warning this week, saying the hoaxers might escalate their efforts as AI makes these attacks easier and more convincing. Read the full AP report →