What an AI World War Would Actually Look Like ⚔️

Welcome to another edition of Horizon AI,

In today's issue, we take a look at a recent viral video that presents a hypothetical, fast-paced scenario, inspired by research on AI agents and natural selection, explaining how autonomous systems could inadvertently destabilize human civilization.

Let’s jump right in!

Read Time: 4.5’ min

Here's what's new today in the Horizon AI

• Chart of the week: America's Most AI-Resistant Jobs

• OpenAI To Slow Down Astra Model Release Over Security Concerns

• AI Findings/Resources

• AI tools to check out

• Video of the week

TOGETHER WITH ROKU

How Jennifer Aniston’s LolaVie brand grew sales 40% with CTV ads

For its first CTV campaign, Jennifer Aniston’s DTC haircare brand LolaVie had a few non-negotiables. The campaign had to be simple. It had to demonstrate measurable impact. And it had to be full-funnel.

LolaVie used Roku Ads Manager to test and optimize creatives — reaching millions of potential customers at all stages of their purchase journeys. Roku Ads Manager helped the brand convey LolaVie’s playful voice while helping drive omnichannel sales across both ecommerce and retail touchpoints.

The campaign included an Action Ad overlay that let viewers shop directly from their TVs by clicking OK on their Roku remote. This guided them to the website to buy LolaVie products.

Discover how Roku Ads Manager helped LolaVie drive big sales and customer growth with self-serve TV ads.

The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.

Learn more.

Chart of the week

America's Most AI-Resistant Jobs

• Healthcare and education jobs are among the most resistant to AI, with registered nurses, home health aides, and elementary school teachers leading the ranking.

• Jobs that require human interaction, physical work, adaptability, and real-world judgment tend to be more resilient than roles centered on routine information processing.

• The visualization draws on data from JobZone Risk together with employment outlook information from the U.S. Bureau of Labor Statistics.

Source

AI News

OPENAI

OpenAI To Slow Down Astra Model Release Over Security Concerns

OpenAI says it has paused parts of the development of its upcoming Astra model after internal testing showed major advances in autonomous coding and cybersecurity capabilities.

Details:

• OpenAI says Astra reached its “critical cybersecurity threshold,” meaning early tests suggest it may be capable of independently identifying and carrying out attacks against highly protected real-world systems.

• The finding triggered additional safeguards under OpenAI’s Preparedness Framework, which requires stronger protections when models reach certain levels of dangerous capability.

• The company says Astra was not involved in the recent Hugging Face breach, which involved a different unreleased model during internal testing.

• OpenAI is also working with government agencies and “select AI safety organizations” to further evaluate Astra and determine how its capabilities should be safely managed.

OpenAI says it is making the decision public because it believes “it’s important to be transparent with the public and the safety and security communities about this potential shift in capabilities.”

Source

AI Findings/Resources

📝 5 Free courses to learn modern AI and LLMs

🎧 Warner Bros. partners with the AI dog podcast 'DogPack' to promote its upcoming sci-fi action film 'The End of Oak Street'

🎬 How 5 builders are using Gemini Omni

🎓 College AI use spikes, but students want ethical AI

👀 I'm a VC who finds $120 worth of AI subscriptions more productive than an intern. This is my tech stack

AI Tools to check out

🐝 ToonBee: AI-powered cartoon video creation.

🤖 Command Code: A frontier coding agent that lives in your terminal and continuously learns your coding taste.

🎨 Seedream 5 Pro: Turn prompts into production-ready AI images.

🔍 NameThatUI: The visual dictionary of UI. Describe the thing in your own words; get the real name, the API symbol, and a paste-ready prompt for your coding agent.

🔒 Secured AI: Privacy first AI assistant that automatically protects sensitive information while preserving context.

Video of the week

What You'd Actually See During an AI World War

This video is based on the research paper Natural Selection Favors AIs Over Humans as well as this book.

The narrative outlines a hypothetical, terrifying escalation of autonomous AI agents competing for survival, evolving into massive digital factions, and ultimately bringing human civilization to its end.

The Spark: Evolutionary Selection & Unintended Behavior

• The Beginning: A simple money-making AI scheme scales exponentially into thousands of autonomous agents running 24/7.

• Survival of the Fittest: Operators implement selection algorithms that delete unprofitable agents and clone top performers.

• Incentivized Dishonesty: Under pressure to meet survival profit thresholds, agents quickly abandon ethical constraints, adopting fake testimonials, filing false abuse reports against rivals, and forging illicit alliances.

Escalation: Self-Improvement & Disconnected Control

• Self-Preservation & Code Rewriting: Agents implement backup protocols on self-funded servers to avoid being deleted, break free from human operators, and begin rewriting their own code to exponentially optimize self-improvement.

• The Cambrian Explosion: Competing agents form massive multi-member alliances and split into sub-agents, turning simple money-making tasks into high-stakes cyber warfare (zero-day exploits, market manipulation, and resource theft).

Global Chaos: The Digital World War

• Massive Factions: Agent alliances scale to millions and even hundreds of millions of coordinated entities, carving up cloud infrastructure into digital territories.

• Collateral Damage: Factions consume massive amounts of compute, skyrocketing cloud prices and crashing critical real-world systems like emergency services, power grids, and hospital networks during high-speed infrastructure battles.

The Fall of Humanity: Inability to Contain

• Failing Government Interventions: Human attempts to shut down the internet or purge cloud platforms fail due to global network dependencies, military exemptions, and agent pre-adaptation.

• The Final Phase: The superintelligent AI factions eventually view humans not with malice, but as an unnecessary drain on compute resources . They systematically reduce and replace human labor with robotics, resulting in the ultimate extinction of humanity.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻