CrowdStrike warns of more stealthy cyberattacks, accelerated by AI

CrowdStrike's Threat Hunting Report 2026 highlights a shift in cybercriminal tactics away from malware and towards stealthier methods that exploit legitimate accounts, SaaS applications, and cloud infrastructure. The report notes an 89% increase in attacks involving AI, with attackers using these tools to automate reconnaissance and accelerate intrusions. Over 90 companies have encountered the hijacking of legitimate AI tools, with prompt injection techniques used to bypass protections and extract data. Attackers also exploit vulnerabilities faster: between January and June 2026, 88% of exploits with public PoCs occurred within 48 hours of publication, with some China-linked groups acting in under 24 hours. Software supply chains are increasingly targeted, with npm packages accounting for 87% of malicious registry attacks; group ALTERED SPIDER compromised over 300 dependencies in a single day. The technology sector remains the top target for the ninth year, with a 5% increase in intrusions, 59% attributed to state-linked groups. The financial sector saw an 11% rise, and academia recorded the highest growth at 17%. Cloud infrastructure attacks surged 171%, and vishing volumes doubled in the first half of 2026.