Roman Strelnikov, head of information security at 1C-Bitrix, warns that the rapid adoption of AI by both attackers and defenders is creating a dangerous arms race in 2026.
Attackers are now automating reconnaissance, vulnerability exploitation, and data exfiltration at speeds previously impossible, while companies increasingly delegate critical tasks involving code, data, and infrastructure to AI agents. This shift introduces new risks that organizations must address urgently.
Risk 1: Speed — AI dramatically accelerates cyberattacks
According to Anthropic's September report, the use of AI is directly linked to faster reconnaissance, exploitation of vulnerabilities, and theft of sensitive data. One prominent example involves the activity cluster designated GTG-20006, which Anthropic attributes to cyber-espionage operations consistent with the publicly known group Midnight Blizzard.
The campaign, referred to in the report as russian espionage, targeted government institutions, diplomatic organizations, and defense industry enterprises. More than 20 organizations were involved in the planning, reconnaissance, and execution phases of this long-running, multi-vector operation.
Attackers leveraged Claude to develop and modify tools, prepare phishing infrastructure, gather intelligence, and orchestrate individual stages of the campaign. This demonstrates how large language models can compress months of traditional attacker work into days or even hours.
