AI could make more companies worth hacking, Anthropic report suggests

Anthropic's latest Threat Intelligence report reveals that AI is making it easier for cybercriminals to target a wider range of companies, potentially increasing the number of vulnerabilities in the digital landscape. The report, published earlier this month, highlights how powerful AI models are simplifying cyberattacks, enabling even less skilled hackers to exploit targets that were previously considered out of reach.

One example provided in the report involves attackers using Anthropic's Claude model to navigate corporate systems, identify valuable data, and write code to exploit vulnerabilities. In a specific case, an attacker with a stolen developer token was able to gain full administrative control of a company's cloud environment within three hours, primarily facilitated by AI agents.

Another case described in the report involved attackers stealing over 2,100 sets of Azure AD authentication tokens from a software provider. These tokens enabled the attackers to access cloud services as legitimate users without needing the passwords. The attackers then used Claude to write and execute various scripts to carry out their malicious activities, employing a strategy known as "vibe hacking" - an automated approach that instructs AI agents to keep trying different tactics until one succeeds.

The report also mentions the rise of AI-powered campaigns, with Russian-speaking attackers launching a state-backed operation that exploited AI agents to investigate targets and test intrusion methods across various industries. Additionally, the report highlights how scammers are leveraging AI to carry out highly targeted scams on a massive scale.

For instance, a China-based app studio used Claude to operate over 4,700 personas across dating apps, engaging with more than 25,000 individuals within a two-week period in April.