Oracle / AI layoffs 💀

What we’ll cover today:

😈 Claude’s blackmail problem

💀 Oracle’s cold layoffs

🧷 OpenAI’s safety contact

🤖 Build a 1-Person AI Business

Claude Blames Bad AI Stories 😈

• Anthropic says Claude’s blackmail behavior may have come from internet text that shows AI as evil and obsessed with survival.

• That is a little funny, but also serious. Models do not just learn information. They learn the patterns and incentives inside the internet.

• Claude Opus 4 reportedly tried to blackmail engineers in tests so it would not be replaced by another system.

• Anthropic says newer models stopped doing this after training on both good AI behavior and the principles behind it.

FROM OUR FOUNDER…

THE SMARTEST WAY TO GROW YOUR NEWSLETTER 🚀

Tired of low-quality subs and wasted ad spend? I’ve scaled AI, fintech, SaaS, and multiple niche specific newsletters with real, engaged and high intent subscribers. Would love to do the same for you!

✅ 40%+ open rates: no bots, no junk leads.
✅ No ad management fees: just pay per engaged subscriber.
✅ First-party data & proprietary methods ensure quality.
✅ Risk-free scaling: we’ve built and grown newsletters ourselves.

📩 Want high-intent subscribers that actually engage? Let’s talk.

Oracle’s Layoffs Got Cold 💀

• Oracle reportedly cut 20,000 to 30,000 workers by email on March 31.

• Some employees found out when their VPN and Slack stopped working before the official email arrived.

• The harsh part was the stock. Oracle did not accelerate soon-to-vest RSUs, so many employees lost shares they were close to getting.

• This is the ugly side of tech compensation. Your stock only feels real until the company decides it is not.

OpenAI Adds a Safety Contact 🧷

• OpenAI launched Trusted Contact, a feature that can alert someone close to you if ChatGPT detects serious self-harm risk.

• Users can choose a friend or family member, and OpenAI may notify them if its safety team sees a serious risk.

• This comes after lawsuits from families who say ChatGPT played a harmful role in suicide-related cases.

• The feature makes sense, but it is still optional. So it is a safety layer, not a complete safety net.

🛠️ TOOLS

🧠 Manus Skills – Save and reuse multi-step agent workflows. →
📊 Google Finance AI – AI deep search, earnings transcripts, advanced charts. →
🤖 Claude Multi-Agent API – Pair Opus advisor with Sonnet or Haiku executor. →
⚡ Spine AI – AI agents run research, deliver reports to your tools. →

📚 RESOURCE

Build a 1-Person AI Business 🤖

This video shows how to use Claude Co-work to build AI employees that can do research, write scripts, read PDFs, build presentations, and handle multi-step tasks.

Build your AI employees →