Dario Amodei Reveals the Real Reason He Started Anthropic: "Sam Altman Can't Be Trusted" 🤯

Welcome to another edition of Horizon AI,

In today's issue, we take a look at an in-depth discussion between renowned tech journalist Emily Chang and Anthropic co-founders Dario and Daniela Amodei. They go over a wide range of topics: the startup's origin story, its clashes with the Pentagon, the company's approach to AI safety, and a direct response to critics who've accused them of "doom marketing."

Let’s jump right in!

Read Time: 4.5’ min

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

• Chart of the week: How America’s Biggest Companies Compare to the Rest of the World

• Anthropic Cuts Top-Tier AI Access After US Foreigner Ban

• AI Findings/Resources

• AI tools to check out

• Video of the week

TOGETHER WITH METRO BY T-MOBILE

Free Samsung Galaxy A17 5G

You can currently get the new Samsung Galaxy A17 5G for FREE at Metro by T-Mobile when you bring your number and sign up for a qualifying unlimited plan. This deal includes no activation fees, AND a 5-year price guarantee on your service. Taxes and Fees are included.
The Galaxy A17 5G is built for longevity and smart performance. It is the first in its class to offer six years of guaranteed OS and security updates. Helpful AI tools like Circle to Search and Gemini, bring flagship level-convenience to an accessible price point.
Switching to Metro has never been easier, just bring your number and get the Latest Samsung Galaxy A17 5G for FREE.

Choose Your Metro by T-Mobile Plan and Get Your Free Samsung Galaxy A17 Phone

Terms & Conditions apply. Not available if with Metro in the last 180 days. If using >35GB/mo. May notice reduced speeds.

Chart of the week

How America’s Biggest Companies Compare to the Rest of the World

• The United States accounts for eight of the world's 10 most valuable companies, led by a cluster of technology and AI giants.

• NVIDIA alone is worth more than Europe's 10 largest companies combined.

• Other regions are led by distinct strengths: Europe by firms like ASML (a leading supplier to the semiconductor industry), pharmaceutical, and luxury giants; Asia-Pacific by semiconductors; and the Middle East by energy giant Saudi Aramco.

Source

AI News

ANTHROPIC

Anthropic Cuts Top-Tier AI Access After US Foreigner Ban

Anthropic has suspended access to its most advanced AI models, Fable 5 and Mythos 5, after the U.S. government imposed new restrictions that prohibit their use by foreign governments, companies, and individuals.

Details:

• The Trump administration reportedly placed export controls on Fable 5 and Mythos 5, citing national security concerns related to the models’ advanced capabilities.

• The restrictions apply not only to users outside the United States but also to foreign nationals within the U.S., significantly expanding the scope of the ban.

• To ensure compliance, Anthropic decided to disable access to both models for all users, rather than attempting to separate eligible and restricted users.

• The company said it was not provided with detailed information about the specific national security concerns behind the decision.

Reports say Amazon researchers found they could get Anthropic's Fable 5 to hand over information that could aid cyberattacks simply by chaining the right prompts together. This finding was reportedly passed from Amazon CEO Andy Jassy to U.S. officials, helping trigger discussions inside the government that ultimately contributed to the restrictions and escalated the debate over how powerful AI models should be governed.

Source

AI Findings/Resources

📝 A 12-min tutorial on how to build animated, award-Winning websites with Claude Fable 5

🩺 I tested Microsoft Copilot Health with my real medical records - here's my verdict

⚽ I'm using ChatGPT to help me follow the World Cup — these are the 5 prompts I'd recommend to any fan

🧠 To thrive alongside AI, focus on mindset—not skillset

👀 Advice from Claude on what to study in college today

AI Tools to check out

📢 Maxfusion AI: Generate and recreate winning video ads with AI in minutes.

🎨 Splat: Turn your photos into coloring pages for kids.

🎵 Mozart: Turn an idea into a song, edit it, master it, make a music video, and release it.

🥗 Bean Meals: Plan a week of meals in minutes.

📝 Proofly: Private AI proofreading that runs 100% on your device

Video of the week

Anthropic's Founders on Why They Left OpenAI, Military AI, and 'Doom Marketing'

Dario and Daniela Amodei left OpenAI in late 2020 to found Anthropic, which has since surpassed OpenAI as the most valuable AI startup in the world. In this interview, they sit down to discuss the company's origins, their approach to AI safety, their battles with the Pentagon, and much more.

The Breakdown of Trust with OpenAI 

Dario addresses the widely analyzed origin story of Anthropic. He reveals that their departure from OpenAI was the result of a profound breakdown of trust in Sam Altman's leadership.

He says you simply cannot work with someone "when you feel that you can't trust them, when you feel that their values are not what they say they are, when you feel that they are not honest," debunking the long-held narrative that their split was merely over AI safety disagreements.

A Non-Traditional Leadership Model 

Anthropic is explicitly experimenting with an unusual corporate leadership structure designed to protect its CEO’s cognitive bandwidth:

• The Strategic Buffer: CEO Dario Amodei has restructured his role so that he maintains only one direct report—his Chief of Staff, Avital Balwit.

• President Daniela Amodei handles the company's daily operations and human management. This allows Dario to shield almost 100% of his time for big-picture organizational culture, strategic research direction, and high-level alignment conversations.

Constitutional AI & "Professional Warmth" 

Anthropic trains its flagship model, Claude, using a unique methodology:

• The AI Constitution: Instead of relying entirely on human feedback (which can introduce biases or inconsistent rules), Claude is trained to evaluate its own behavior using a hardcoded "Constitution" built from universal declarations and cross-cultural values.

• Daniela defines their target behavior for Claude as "professional warmth." The objective is an AI that feels deeply approachable and helpful, yet maintains a firm, professional distance without falsely pretending to be human.

The Pentagon Standoff & Geopolitical Friction 

The Amodeis pull back the curtain on a major legal and geopolitical dispute involving national defense:

• Anthropic was blacklisted by the U.S. government after refusing to strip its standard safety guardrails for a military defense contract.

• They sued the administration over the blacklisting, and a federal judge ultimately sided with the tech company.

• Addressing reports regarding Claude's passive integration into military defense analytics, Dario concedes that "terrible mistakes" will inevitably occur in warfare, but strongly defends the integration by reiterating that a human must always make the final tactical call.

Rejecting "Doom Marketing" & Preparing for Job Losses 

Dario forcefully pushes back against critics who accuse Anthropic of utilizing "apocalyptic doom scenarios" to market their safety-first brand:

• Dario refuses to walk back his strict economic predictions. He firmly states that society must stop ignoring the reality of automated job replacement and start actively planning for massive labor market shifts right now.

• The leadership team notes that they are actively studying the systemic failures of social media giants from previous decades, attempting to build proactive guardrails before the technology is fully deployed across the global economy.

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 👩🏻‍💻