Anthropic got hacked | Eli the Computer Guy
Nov 2, 2026 · 31m
More Technology podcasts · More News podcasts
Summary
Host Isaac and guest Eli the Computer Guy discuss a supply chain attack where a compromised Anthropic extension on Cursor installed malicious SSH backdoors on up to 5,000 developer machines. The conversation highlights the severe security incompetence of frontier AI labs, contrasting their existential risk rhetoric with poor infrastructure practices. The hosts also analyze the growing trend of companies like Amazon blocking external AI agents to prevent resource exhaustion and liability, arguing that agentic commerce is currently too risky and inefficient for widespread adoption.
Topics discussed
Sponsors: Google Workspace and PreNuvo
Intro: Anthropic security failures
The Anthropic/Cursor supply chain hack
Explaining IDEs, extensions, and the compromise
SSH hijacking and root-level risks
Company response and PR implications
Anthropic's lack of basic logging
Risks of agents in core infrastructure
Sponsors: Google, PreNuvo, and Toast
Amazon and banks blocking AI agents
The API wars and agent resource costs
Agent inefficiency and e-commerce liability
Walled gardens and internal agent development
Competitive dynamics: Amazon vs Meta
Resource usage: Agents vs Humans
Sponsors: Google, PreNuvo, and Toast
Dynamic pricing and server crashes
Liability, scams, and AEO risks
AI as a heads-up vs full automation
Infrastructure limits and MCP protocols
Outro and Sponsors: Camtasia and SkyRizzy
More from The Tech Report
Listen ad-free on Castria