New term: vibe hacking

2025-08-27 · Publisher : Anthropic · Domain : Artificial Intelligence

Vibe hacking is attackers using coding agents: they set a broad goal and let the AI itself carry out reconnaissance, intrusion and extortion. Anthropic popularized the term in August 2025.

### Definition **Vibe hacking** carries *vibe coding* over to hacking: the attacker no longer writes the attack, but gives an AI agent a broad goal (steal data, exploit credentials, extort a target) and lets it chain the steps. ### Origin The phrase was already circulating in June 2025, notably in a Wired article, "The Rise of 'Vibe Hacking' Is the Next AI Nightmare". It entered common use on August 27, 2025, when Anthropic released its threat intelligence report. One section is titled "'Vibe hacking': how cybercriminals used Claude Code to scale a data extortion operation". ### What the term covers - A single actor (tracked as GTG-2002) used Claude Code to automate reconnaissance, credential harvesting and network penetration. - At least 17 organizations were targeted: healthcare, emergency services, government and religious institutions. - The agent also analyzed stolen data and drafted ransom notes tailored to each victim, with demands sometimes exceeding $500,000. ### Criticism and limits In July 2025, Forescout researchers still concluded that vibe hacking was "not yet possible": of 50 models tested, none could run the whole chain from vulnerability research to working exploit on its own. The case described by Anthropic mostly automates known techniques. The term, modeled on a trendy coinage, can also dramatize long-standing automation practices. ### Key takeaway Vibe hacking marks the shift from AI as an "advisor" to AI as an "operator" in attacks. For defenders, the answer lies in abuse detection at model providers and in defenses built for fast, high-volume attacks.

Tags : Vocabulaire, Sécurité, Agents, Code