Anthropic Says Most Banned Accounts Used AI to Prepare Cyberattacks
Anthropic said about 67% of banned accounts it reviewed used AI to help prepare for cyberattacks. The finding points to a growing security challenge as malicious actors become more capable with AI tools.
What happened?
Anthropic said about 67% of banned accounts it reviewed used AI to help prepare for cyberattacks. The finding points to a growing security challenge as malicious actors become more capable with AI tools.
Why it matters
AI firm Anthropic said about 67% of the banned accounts it analyzed used AI to prepare for cyberattacks, according to source material from Cointelegraph. The company mapped a year’s worth of AI-enabled cyber threats and found that malicious actors are becoming more dangerous as they adopt AI tools.
AI firm Anthropic said about 67% of the banned accounts it analyzed used AI to prepare for cyberattacks, according to source material from Cointelegraph. The company mapped a year’s worth of AI-enabled cyber threats and found that malicious actors are becoming more dangerous as they adopt AI tools.
The finding matters because cyber risk is already a core concern for crypto users, companies and infrastructure providers. If attackers can use AI to speed up preparation or improve the quality of their tactics, exchanges, wallet providers, protocols and individual users may face a more difficult threat environment.
Anthropic’s review focused on accounts it had banned, meaning the figures describe activity detected within that enforcement set rather than the entire internet or all AI usage. Still, the reported share highlights how quickly AI can become part of malicious workflows when access controls and monitoring fail to stop abuse early.
For the crypto sector, the report adds to broader concerns around phishing, social engineering and attacks on digital infrastructure. The source material does not identify specific crypto targets, prices or incidents, but the security implications are relevant for an industry where stolen credentials and compromised systems can lead directly to asset losses.
The takeaway is not that AI is inherently malicious, but that its misuse is becoming a practical security issue. Companies using or providing AI tools will likely face continued pressure to detect abusive behavior while legitimate users rely on stronger operational security to limit exposure.