Anthropic report reveals AI increasingly used in cyberattack preparation

Artificial intelligence is becoming an increasingly important tool for cybercriminals, according to a new analysis by AI company Anthropic, which found that approximately 67 per cent of banned accounts on its platform had used AI systems to assist with preparations for cyberattacks. The findings highlight how rapidly threat actors are adopting advanced AI technologies, raising new concerns about digital security and the evolving nature of cybercrime.

AI becoming a tool for cyber operations

Anthropic’s research examined a year’s worth of AI-enabled security threats and abuse cases identified on its platform. The company found that while many malicious actors continue to rely on traditional hacking techniques, artificial intelligence is increasingly being incorporated into planning, reconnaissance and operational support activities.

Rather than carrying out attacks directly, AI was most frequently used to accelerate preparatory tasks, helping threat actors gather information, automate research and improve the efficiency of their operations.

Security experts note that these developments reflect a broader trend in which cybercriminals are integrating emerging technologies into existing attack methods.

Preparation rather than execution

The report suggests that AI is currently most valuable to attackers during the planning phase. Threat actors reportedly use AI systems to analyse technical information, identify vulnerabilities, generate code, create phishing content and research potential targets.

By reducing the time required for these activities, AI can enable both experienced cybercriminals and less technically skilled actors to operate more effectively.

Researchers emphasise that while AI does not eliminate the need for human expertise, it significantly lowers barriers to entry for certain forms of cyber activity.

Threat landscape evolving rapidly

According to Anthropic, malicious actors are adapting quickly as AI capabilities improve. Cybersecurity professionals warn that future AI systems could further enhance the speed, scale and sophistication of attacks if adequate safeguards are not implemented.

The findings have renewed debate about how AI companies should monitor platform usage and prevent misuse without restricting legitimate applications of the technology.

Major AI developers are investing heavily in safety systems designed to identify suspicious behaviour, block harmful requests and remove accounts that violate usage policies.

Growing focus on AI safety

Anthropic has positioned AI safety as a central element of its business strategy. The company regularly publishes research into misuse patterns and collaborates with cybersecurity experts to strengthen defensive measures.

Industry observers view transparency regarding abuse attempts as increasingly important as AI adoption expands across governments, businesses and critical infrastructure operators.

The challenge facing developers is balancing accessibility and innovation with effective safeguards against malicious activity.

Defenders are using AI too

While much attention focuses on how criminals use artificial intelligence, cybersecurity professionals are also deploying AI-powered tools to improve threat detection, automate incident response and identify vulnerabilities before they can be exploited.

Many experts believe the future of cybersecurity will involve an ongoing technological competition in which both attackers and defenders increasingly rely on AI-driven capabilities.

The outcome will depend largely on which side can adapt more quickly to emerging technologies.

A new chapter in cyber risk

The report highlights how artificial intelligence is reshaping the cybersecurity landscape. As AI systems become more powerful and widely available, organisations will need to strengthen security practices and prepare for increasingly sophisticated threats.

Although the technology offers enormous economic and societal benefits, Anthropic’s findings demonstrate that malicious actors are already exploring ways to leverage those same capabilities for harmful purposes.

For businesses, governments and technology providers, the message is clear: AI is becoming a critical factor in cybersecurity, and the race to manage its risks has only just begun.

Newshub Editorial in North America – 4 June 2026