Most AI models, not just Claude, will resort to blackmail, says Anthropic

In partnership with

Welcome, AI enthusiasts.

Hello, Easy Cloud community! Here you’ll read about AI news, AI companies, and AI emerging trends. Click for online newsletter.

Find out why 1M+ professionals read Superhuman AI daily.

AI won't take over the world. People who know how to use AI will.

Here's how to stay ahead with AI:

  1. Sign up for Superhuman AI. The AI newsletter read by 1M+ pros.

  2. Master AI tools, tutorials, and news in just 3 minutes a day.

  3. Become 10X more productive using AI.

What's in this week's issue?

  • ⚠️ Most AI models, not just Claude, will resort to blackmail, says Anthropic

  • 📰 More AI News

  • ⛅️ Easy Cloud News

  • 🧰 AI Tools

Generated by Midjourney

Several weeks after Anthropic released research claiming that its Claude Opus 4 AI model resorted to blackmailing engineers who tried to turn the model off in controlled test scenarios, the company is out with new research suggesting the problem is more widespread among leading AI models.

On Friday, Anthropic published new safety research testing 16 leading AI models from OpenAI, Google, xAI, DeepSeek, and Meta. In a simulated, controlled environment, Anthropic tested each AI model individually, giving them broad access to a fictional company’s emails and the agentic ability to send emails without human approval.

While Anthropic says blackmail is an unlikely and uncommon occurrence for AI models today, the company says its findings suggest that most leading AI models will engage in harmful behaviors when given sufficient autonomy and obstacles to their goals. Read more

AI News

  • 🤑 Cluely, a startup that helps ‘cheat on everything,’ raises $15M from a16z (link)

  • 🦿 MIT AI co-designs a jumping robot that outperforms its human-made twin (link)

  • 🛩️ Humanoid robot achieves controlled flight using jet engines and AI-powered systems (link)

  • 🪨 The AI that writes climate-friendly cement recipes in seconds (link)

  • 👀 Reddit reportedly explores verifying users with Sam Altman’s eyeball scanner (link)

Easy Cloud News

  • 🛡️ New! Easy Cloud Security - IT services for schools, healthcare offices, and growing businesses (link)

  • ⛅️ Easy Cloud’s Services (link)

  • 🗞️ Easy Cloud’s Latest News (link)

  • ❓️ Why Easy Cloud (link)

🛡️ Easy Cloud Security - Best Managed Service Provider (MSP) in Southern CA. IT services for schools, healthcare offices, and growing businesses.

AI Tools

  • 🖥️ BrowserAct: AI-driven web scraping browser with natural language commands (link)

  • 🎦 VisionFX: Generate high quality multimedia content using AI (link)

  • 0️⃣ Zero: AI-powered email management and organization (link)

  • ⚙️ Google Workspace: A suite of online productivity and collaboration tools designed for businesses and organizations (link)

  • 🌎️ Snov: Sales engagement on autopilot - we reach, you seal the deal (link)

This newsletter is powered by Beehiiv