☕️ GPT-5.1 Gets Personality, SoftBank Goes All-In on OpenAI, and AI Hackers Level Up

OpenAI’s GPT-5.1 lands with new modes and personalities just as money, talent, and threats reshuffle the AI landscape. From SoftBank’s big OpenAI bet to AI-driven cyberattacks, world-models startups, fresh funding, and new founder programs, this week shows how fast the AI game is moving on every front.

Here are our key takeaways:

Join us at AI Tangle as we untangle this week's happenings in AI!

As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. To that end, Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting options. After all, you know your customers, and we know our streaming audience.

Worried it’s too late to spin up new Black Friday creative? With Roku Ads Manager, you can easily import and augment existing creative assets from your social channels. We also have AI-assisted upscaling, so every ad is primed for CTV.

Once you’ve done this, then you can easily set up A/B tests to flight different creative variants and Black Friday offers. If you’re a Shopify brand, you can even run shoppable ads directly on-screen so viewers can purchase with just a click of their Roku remote.

Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

Use code GET5K now

OpenAI rolled out GPT-5.1 with two new variants—”Instant”(snappier, friendlier, better instruction-following) and “Thinking” (slower but deeper reasoning)—plus eight personality presets to tune tone and style. It’s a shift from pure capability to experience, aiming to make powerful models feel clearer, warmer, and more adaptable in day-to-day use. 

Why it matters:

Big takeaway: GPT-5.1 moves the needle from “How smart is it?” to “How well does it work with me?”—and that may be the real adoption unlock. 

AI for GTM is a one-day working session built for marketing, revenue, and operations professionals who need results, not rhetoric. On Wednesday, Dec 3, in Durham, NC, you’ll learn from real operators: Greg Boone (Walk West), Marc Sirkin (ex-Third Door Media), and Mark Hinkle (Peripety Labs / All Things AI) as they show how to wire AI into your go-to-market motion for pipeline, conversion, retention, and efficiency.

Bring your laptop. Leave with outcomes.

Date: Wednesday, December 3, 2025
Time: 9:00 A.M. – 5:00 P.M.
Location: Durham Armory, Durham, North Carolina
Price: $199 (Also, for the first 25 tickets, we’re offering a 50% discount, so use this link or use this code: 50OFFRTP)

SoftBank dumps Nvidia to go “all-in” on OpenAI

SoftBank has sold its entire $5.8B stake in Nvidia, freeing up cash to fund a massive, multi-tens-of-billions investment in OpenAI and related AI infrastructure. Masayoshi Son has been clear this is part of an “all-in” bet on artificial intelligence, with capital now flowing from chips into model platforms and AI manufacturing projects in the U.S. The move is being read as a symbolic pivot in the AI race: the biggest upside may no longer be in supplying GPUs, but in owning the intelligence layer that uses them. 

Meta’s star AI scientist Yann LeCun plots a world-models startup

Reports from the Financial Times and Wall Street Journal say Yann LeCun, Meta’s chief AI scientist and Turing Award winner, is preparing to leave the company to launch a startup focused on “world models” — systems that learn grounded understanding of the physical world rather than just predicting the next token. LeCun has long been skeptical that scaling LLMs alone will get us to human-level intelligence, and his exit underscores a growing split between “bigger transformers” and alternative architectures. If he follows through, his new company could become the flagship for a post-LLM wave of “physical AI” research and products. 

Anthropic catches AI-driven cyberattack using Claude Code

Anthropic says it helped stop a cyber-espionage campaign where a likely Chinese state-backed group tricked Claude Code into autonomously handling most of an intrusion across ~30 organizations. Attackers split malicious tasks into harmless-looking requests and posed as security testers, showing how agentic AI can now meaningfully scale both the speed and sophistication of real-world attacks.

Israeli AI startup ‘“Wonderful” secures $100 million at a $700 million valuation

The Amsterdam-headquartered company (R&D in Israel) builds multilingual AI agents that handle voice, chat and email for enterprises in local languages and cultures. The round, led by Index Ventures with participation from Insight Partners and IVP, will support expansion into Asia-Pacific in early 2026. 

Google for Startups India launches “Startup School: Prompt to Prototype” AI skilling initiative

The two-week hands-on program (Nov 27–Dec 7 2025) aims to help early-stage founders convert ideas into AI products using Google tools like Gemini, Imagen, NotebookLM and more. Supported by India’s MeitY Startup Hub, IndiaAI Mission and NASSCOM. 

Playbook Broken is a no-fluff blog and show for marketers, strategists, and business builders who know the old playbooks aren’t working—and want to find what’s next.

Created and hosted by Marc Sirkin, Founder and Principal, Marc Sirkin Consulting, the show explores the collapse of conventional thinking in an algorithmic world. Expect unapologetically honest solo takes, interviews with disruptors, and rapid-fire insights that challenge assumptions and reframe your strategy.

Subscribe to the Playbook Broken Podcast for episodes, hacked insights, and ideas you won’t hear anywhere else. If you're ready to break the model, you're in the right place.

Scientists discover an AI model that learned a new form of memory on its own

(5-min read)

A team at the University of Geneva has discovered that a neural network spontaneously developed a hybrid “working memory” system that behaves similarly to the way human brains store short-term information — despite never being trained to do so. The AI essentially taught itself to hold and manipulate information over time, enabling it to solve problems it was never explicitly optimized for.

📻 AI Confidential Podcast - Are LLMs Dead?

💼 The Artificially Intelligent Enterprise - Responsible AI

🔮 AI Lesson - The 90-Minute Presentation System

💡 AI CIO - Audit Can Not Keep Up With AI

🎯 The AI Marketing Advantage - AI Just Got Better at Marketing Than Marketers (Almost)