This is a transcript of the Artificial Intelligence Show podcast, hosted by Paul Ritzer and Mike Kaput.

Introduction

"This BS stuff that's coming from leaders of the labs and politicians... it is disingenuous and it is harmful to take this highly confident belief that it'll just work out. You do not know that. No one knows that. So at minimum we should be doing contingency planning for worst case scenario outcomes... You cannot pretend like you understand the future because you're an economist or you're an AI leader or whatever it is and that like it just works out. You don't know that."

Welcome to the Artificial Intelligence Show, the podcast that helps your business grow smarter by making AI approachable and actionable. Each week, we break down the AI news that matters and provide insights to advance your company and career.

This week's episode, recorded on March 16th, explores the escalating conflict between Anthropic and the Pentagon, the controversy around a New York Times AI writing quiz, and the growing trend of AI-driven job cuts.

AI Pulse Survey Results

Each week, we conduct an informal AI pulse survey with our listeners. With nearly 70 responses last week, here are the results:

Main Topics

Anthropic vs. Pentagon: Round 3

The conflict between Anthropic and the Pentagon continues to escalate. After the Pentagon labeled Anthropic a supply chain risk—a designation usually for foreign adversaries—the story has moved into the legal arena.

Anthropic has filed two federal lawsuits to block this blacklisting and sought an emergency temporary restraining order. The company’s CFO stated that hundreds of millions of dollars in expected revenue are already at risk, with the potential for billions in losses if enterprise customers follow the Pentagon's lead. They've already reported a lost $100 million deal and disruptions to negotiations worth over $180 million.

Support for Anthropic has come from several influential sources:

Pentagon CTO Emil Michael claimed on CNBC that Anthropic's model, Claude, would "pollute the defense supply chain" because its safety features make it unreliable for military operations. He pointed to the model having a "soul and a constitution" that is not the US Constitution, and even referenced a belief that it has a 20% chance of being sentient.

However, Palantir CEO Alex Karp confirmed that Claude is still actively used in military operations because there is no viable replacement. Analyst Dean Ball, former lead author of the Trump administration's AI action plan, pointed out that all frontier AI models have similar behavioral guidelines. If Anthropic's guardrails are a risk, the same logic could apply to any AI provider.

This situation appears to be less about technology and more about politics and control. The administration seems to dislike Anthropic's principles and wants more control over how the models behave. With powerful investors like Google, Microsoft, Amazon, and Nvidia, and a planned IPO in 2026, Anthropic is a major player. The situation is likely too intertwined with influential figures on both sides of the political aisle for the government to continue its current path without severe blowback. A negotiated deal seems to be the only viable off-ramp for all parties involved.

New York Times Releases Controversial "AI Writing Quality" Quiz

The New York Times published an interactive quiz asking readers, "Can you tell the difference between AI-generated writing and human-written prose?" The quiz presented short passages—one by a human author, one by an AI—and asked readers to identify the AI and state their preference.

The results were striking:

The methodology drew immediate criticism from writers who argued that short, decontextualized fragments do not represent the craft of writing, which involves sustained voice, structure, and argument. However, supporters claimed the results are significant, suggesting the gap in surface-level writing quality has effectively closed.

This debate highlights a fundamental question: not if AI can write well, but when we should use it. This decision exists on a spectrum and is a personal choice. A useful framework is the Human-to-Machine Scale for Writers:

The choice depends on the need for authenticity, emotional nuance, and unique perspective. When the process of writing is the purpose, the human should lead. When the value is in the information itself and speed is critical, the machine can take over.

Atlassian Layoffs and the AI Job Loss Dashboard

Atlassian, the company behind Jira and Trello, announced it is cutting approximately 1,600 jobs (10% of its workforce), explicitly attributing the move to preparing for the "AI era." This is one of the first times a major tech company has publicly named AI as the primary driver for large-scale layoffs, rather than citing broader economic conditions.

The cuts come despite strong growth, with cloud revenue up 26% year-over-year. The company's stock rose on the news. More than half of the layoffs were engineers, a reversal from the CEO's prediction 18 months ago that AI would lead to hiring more engineers.

Related to this, the Alliance for Secure AI launched jobloss.ai, a live dashboard tracking AI-linked job losses. As of this week, it reports over 76,000 total losses globally since January 2025.

This trend is creating significant uncertainty. Bill McDermott, CEO of ServiceNow, predicted that unemployment for new college graduates could reach the mid-30s in the next few years as AI agents take over entry-level work.

In response to this growing concern, organizations like the Windfall Trust are emerging. Spun out of the Future of Life Institute, the Windfall Trust aims to build a global social safety net to alleviate the economic impact of AI-driven joblessness. They advocate for scenario planning, a standard practice in national security, to prepare for various outcomes of AI's impact on the economy. The core message is that uncertainty is not an excuse for inaction; it is a reason to plan smarter for all contingencies, including worst-case scenarios.

Rapid Fire Topics

Adobe CEO Stepping Down Amid AI Pressure

Shantanu Narayen, Adobe's CEO for the past 18 years, announced he will step down. This comes as Adobe's stock is down significantly, and the company faces intense pressure from generative AI competitors eroding demand for its core creative suite. While Q1 earnings beat expectations, with AI-related revenue tripling, there is a growing divide between positive public statements from executives and the competitive risks acknowledged in private SEC filings. Many software companies are now citing AI agents as a major competitive threat that could allow customers to replicate their apps or draw data from them.

Amazon's AI-Related Outages and Engineering Struggles

Amazon experienced four high-severity incidents in a single week, one of which was traced to a novel failure: a human engineer consulted an AI agent that provided inaccurate troubleshooting advice based on an outdated internal wiki. This triggered a cascade of failures. A Guardian investigation also found that Amazon engineers are being heavily pressured to use internal AI coding tools that frequently generate unreliable code, forcing them to spend more time fixing the AI's mistakes. This highlights the risks of rushing AI adoption without proper governance, training, and change management.

McKinsey's AI Chatbot Hacked by an Autonomous Agent

A security startup named Codewall revealed its autonomous AI agent breached McKinsey's internal AI chatbot, "Lily," in under two hours. The agent gained read-access to 46.5 million chat messages, 57,000 user accounts, and confidential client data file names. The incident underscores a critical vulnerability for enterprises: the real risk often lies not in the AI model itself, but in the weakly governed APIs and internal services that AI agents can access and manipulate.

AI Politics Update

The political lines on AI are becoming clearer. Senator Bernie Sanders called for a federal moratorium on the construction of new AI data centers, citing the strain on power grids and environmental damage. Meanwhile, the Trump administration is working to override a patchwork of state-level AI regulations in favor of a lighter-touch federal framework that prioritizes industry growth over prescriptive rules.

Grammarly's "Expert Review" Controversy

Grammarly faced swift backlash after it was revealed the company had commercially deployed AI clones of public figures like Stephen King and Kara Swisher to sell a premium "Expert Review" feature without their permission. The feature used AI models based on the experts' public writing to mimic personal feedback. Following a class-action lawsuit and public outcry, Grammarly killed the feature and apologized, acknowledging its "take first, apologize later" approach was a fundamental mistake.

Andrej Karpathy's Autonomous AI Researcher

Former OpenAI co-founder Andrej Karpathy shared results from "Auto Research," a Python script that autonomously runs AI research experiments. Left running for two days, the script independently discovered and implemented approximately 20 distinct changes that improved a small language model's performance by 11%—a significant leap in AI research. This demonstrates the power of autonomous agents to accelerate discovery and also highlights the risk of "intelligence brownouts," where over-reliance on these systems can lead to catastrophic failure if they stop working.

AI Product and Funding Updates