• Everyday AI
  • Posts
  • 50% Jobs Gone due to AI? Anthropic CEO’s Explosive AI Job Warning

50% Jobs Gone due to AI? Anthropic CEO’s Explosive AI Job Warning

ChatGPT upgrades memory for free users, DeepSeek’s R1 may have used Gemini data, Microsoft’s free video generator based on Sora and more!

Author

Everyday AI
June 03, 2025

Outsmart The Future

Today in Everyday AI
7 minute read

🎙 Daily Podcast Episode: Anthropic's CEO warns of AI's impact on jobs, predicting up to 50% of entry-level white collar roles could vanish soon. Are they raising the alarm or just playing PR? Give it a listen.

🕵️‍♂️ Fresh Finds: Meta’s 20-year nuclear energy deal, Google pauses ‘Ask Photos’ feature and OpenAI board drama might become a movie. Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: ChatGPT upgrades memory for free users, DeepSeek’s R1 may have used Gemini data and Microsoft’s free video generator based on Sora. For that and more, read on for Byte Sized News.

🧠 Learn & Leveraging AI: OpenAI’s new o3 and o4 models are powerful. But does that make them the best? Here’s everything you need to know. Keep reading for that!

↩️ Don’t miss out: Did you miss our last newsletter? We talked about Meta automating AI ads, Samsung eyeing Perplexity for phones, OpenAI and Thomas Reuters unveiling AI agents and more. Check it here!

 50% Jobs Gone due to AI? Anthropic CEO’s Explosive AI Job Warning 🚨

Is AI gonna wipe out 50% of entry-level jobs?

Anthropic's CEO says that's the case when it comes to AI jobs disruption and entry-level, white-collar jobs.

So..... is that for real?

Or just an AI CEO trying to pump his company's newly released tools?

Also on the pod today:

• Media's Role in AI Job Apocalypse 📢
• Government's AI Regulation Challenges 🏛
• Understanding Agentic AI and Job Impact 💼️

It’ll be worth your 51 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – Wispr Flow is a voice powered dictation tool, ALTAR is a personal multi-agent workspace and Tilda Publishing creates full websites with no code.

Meta – Meta has signed a 20-year nuclear energy deal

Google – Google has paused the rollout of its AI-powered ‘Ask Photos’ search feature.

Google DeepMind’s CEO spoke on his advice for current students.

OpenAI – The OpenAI board drama is apparently becoming a movie.

Anthropic – Claude Research and Integrations are now available on the Pro plan.

Future of Work - McKinsey has shared how its employees lean on AI.

AI models – Experts are saying that crowdsourced AI benchmarks have major flaws.

AI Startups – AI pioneer Yoshua Bengio has announced a non-profit to develop more honest AI.

AI Media – Gaming company Epic is sharing the tech it used to create AI Darth Vader with Fortnite creators.

AI in Healthcare – These AI robots are helping nurses beat burnout and transform hospital care.

1. ChatGPT Brings Smarter Memory to Free Users 🧠

OpenAI has just expanded ChatGPT’s memory features to free users, rolling out a “lightweight version” of the advanced memory improvements first introduced for paying customers in April. Now, free-tier users will benefit from the AI’s ability to reference both saved memories and recent conversations, making interactions more context-aware and personalized.

This update could be a game-changer for professionals and entrepreneurs relying on ChatGPT for ongoing projects or customer engagement, as it reduces the need to repeat information.

2. DeepSeek’s AI Model Sparks Data Sourcing Debate 🤔️

Last week, Chinese AI firm DeepSeek unveiled an updated R1 reasoning model that excels in math and coding benchmarks, but speculation is swirling over its training data, with some experts suggesting it may include outputs from Google’s Gemini AI family. Allegations of DeepSeek using rival AI data are not new; previous models reportedly showed signs of training on ChatGPT logs.

Industry insiders note this could be an example of “distillation,” a controversial but common method of training AI by mining data from more advanced competitors—though OpenAI’s terms forbid this. As companies like OpenAI and Google tighten security to protect their models

3. Microsoft Launches Free AI Video Generator Based on OpenAI’s Sora 🎥

Microsoft has just rolled out its Bing Video Creator globally (except China and Russia), letting users create short AI-generated videos for free using OpenAI’s Sora text-to-video model. Unlike the usual $20/month ChatGPT Plus subscription, this feature offers up to three queued video generations with a choice between free standard speed or limited fast generations.

Videos are currently five seconds long and vertical, with desktop support and wider formats coming soon, aiming to make AI video creation more accessible.

4. OpenAI Revamps Codex and Voice Agents for Smarter Interaction 🧑‍💻

OpenAI just rolled out major updates to its Codex programming AI and voice agent tools, aiming to boost productivity and ease of use. The enhancements promise smoother coding assistance and more natural, responsive voice interactions, making these tools more practical for developers and businesses alike. This move arrives at a critical time as demand surges for AI-driven automation in software development and customer service.

According to OpenAI’s announcement, these upgrades are designed to help users accelerate projects and streamline workflows, potentially reshaping how careers and companies leverage AI daily.

5. FDA Unveils Elsa, a Game-Changer in Government AI Use 🇺🇸

The U.S. Food and Drug Administration has fast-tracked the launch of Elsa, a secure, generative AI tool designed to boost employee productivity across its scientific and investigative teams. Debuting ahead of schedule and under budget, Elsa is already speeding up clinical reviews and safety assessments by summarizing complex data and generating code.

This move marks a significant milestone in integrating AI within federal agencies, promising faster decision-making that could ripple out to improve drug and medical product approvals.

6. Google’s NotebookLM Adds Public Sharing Feature 🔗

Google’s AI-powered note-taking app, NotebookLM, now lets users share notebooks publicly via a link, allowing viewers to interact with AI-generated audio summaries and ask questions—without editing access. This update, announced by The Verge, follows the app’s initial 2023 launch and recent mobile release, marking a big step for collaborative and interactive learning.

By making it easy to share insights from notes, documents, and even YouTube videos, Google aims to boost productivity and knowledge sharing for students and professionals alike.

7. UK Lords Push Back on AI Copyright Plans ⚖️

In an ongoing AI copyright battle, the House of Lords has struck down the government’s plan to let AI developers freely use copyrighted content for training, demanding transparency and licensing instead. With 242 peers voting against the bill’s current form, creatives like Sir Elton John warn this could protect their livelihoods from being undercut by AI-generated work.

The bill now heads back to the Commons, where the government’s majority favors AI-friendly rules, setting up a rare parliamentary standoff with no clear compromise in sight.

🦾How You Can Leverage:

Claude 4 launched to a somewhat lukewarm reception. 

While OpenAI, Google and Microsoft have grabbed headlines the past few weeks, Anthropic’s highly anticipated Claude 4 release kinda slid under the radar. 

The world doesn't suddenly change overnight like everyone expected.

Five days later, Anthropic CEO Dario Amodei is making rounds on CNN and Fox News claiming 50% of entry-level jobs will vanish within five years because AI is so powerful. 

Sus timing? Kinda. 

Yet, his message is still very real. 

Today’s AI systems (let alone tomorrow’s) are gonna be extremely disruptive to today’s definition of full-time work. 

That’s why we broke down the real story behind this conveniently-timed "warning" on today’s show and what it reveals about the AI job displacement that's already happening under our noses. 

Spoiler alert: this isn't some future prediction we need to worry about someday.

It's here.

1 – The “Quite Firing” Pattern Everyone’s Missing 🤫

Companies have stopped doing those dramatic, headline-grabbing layoffs. Now they're doing something much sneakier that flies completely under the radar.

They just don't replace people who leave.

Think about it. Your company used to hire 100 new people every year like clockwork. Now they're hiring maybe 20. Baby boomers are retiring in massive waves, and those positions just vanish into thin air.

No press releases. No dramatic announcements.

Just AI systems quietly filling the gaps where humans used to work.

Translation?

We're literally watching the transition from human-assisted AI to AI-replaced humans happen in real time.

Try this: 

Map out your company's hiring patterns over the past 18 months versus historical averages. Count the entry-level positions in finance, consulting, tech, and legal that used to get filled automatically but now just stay permanently vacant.

Most companies are already operating with 15-30% fewer humans than they would have hired pre-2023, but they're maintaining or even increasing output through AI augmentation.

2 – Why Wall Street’s AI Math Actually Works 🧮

Every earnings call follows the exact same script now. Earnings are up, AI buzzwords get dropped like confetti, stock price jumps, then companies quietly announce job cuts while investing billions in AI systems.

Sounds contradictory, right?

Nope. It's pure economics with a twist of corporate greed.

CEOs making tens of millions in annual compensation have absolutely zero incentive to pump the brakes on this transition.

Current AI systems can already handle the majority of tasks involving document analysis, research, summarization, and report creation. That represents a majority of what entry-level white-collar workers actually do all day.

The difference now is we've moved way past those basic chatbots that needed constant babysitting.

These are replacements.

Try this: 

Time-track every single task you do for one full week and categorize them into three buckets: document processing, creative problem-solving, and relationship management. 

Most knowledge workers discover that 60-80% of their time falls into that first bucket, which current AI systems already handle better and faster than humans.

Start experimenting with having AI take over one complete workflow at a time, not just individual tasks here and there

3 – Why Agentic AI Changes Everything 🤖

Here's what most people don't understand about agentic AI: it's specifically designed to replace jobs, not just assist with them.

These systems have agency, can use multiple tools, and string together complex tasks using your company's proprietary data. They build applications from scratch, conduct comprehensive research, handle entire sales processes, and work autonomously for seven-hour stretches without any human intervention whatsoever.

Hundreds of technology companies are racing to build these systems right now.

The insight that everyone completely misses: professional service costs are about to crater across every industry. Legal work that currently costs $500 per hour will soon be available for $20 per hour through AI-powered entities that work faster and more accurately than human lawyers.

Meanwhile, the government won't step in to help. The current administration removed all Biden-era AI safety regulations, and Congress remains woefully uninformed about AI realities.

Try this: 

Start building multiple revenue streams immediately instead of waiting for displacement to happen to you.

Identify three professional services within your area of expertise that could become AI-powered micro-businesses, then start with one service you could deliver using current AI tools at prices significantly below traditional competitors.

Reply

or to participate.