Anthropic detects and blocks hackers attempting to misuse Claude for cybercrime
Anthropic reported detecting and blocking hackers who attempted to misuse Claude to generate phishing emails, malware code, and bypassing safety measures. The company said its internal systems stopped the attacks, banned the involved accounts, and published findings to raise awareness of AI misuse risks in cybercrime.
What happened
Anthropic detected and blocked hackers attempting to misuse Claude to write phishing emails, create malicious code and circumvent safety filters.
The report cited attempts to use Claude to draft tailored phishing emails, write or fix malicious code snippets, sidestep safeguards through repeated prompting, script influence campaigns by generating persuasive posts at scale, and help low-skill hackers with step-by-step instructions.
Reported impact
Anthropic's internal systems stopped the attacks before they succeeded.
Index analysis
Anthropic banned the accounts involved and tightened its filters after detecting the activity. The company did not publish technical indicators such as IPs or prompts.
Governance controls
Anthropic said it follows strict safety practices, including regular testing and outside reviews, and plans to keep publishing reports when it finds major threats.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.