Check Point Software Technologies has released its Annual AI Security Report 2026, warning that artificial intelligence is shifting from supporting cyber criminals to directly executing parts of cyber intrusions, compressing the time defenders have to respond.
The report, produced by Check Point Research, argues that defenders are increasingly contending with “machine-speed” activity rather than human-paced operations, with AI accelerating exploit development, automating intrusion workflows and lowering barriers to entry for attackers.
According to the report’s findings, researchers observed intrusions where AI ran exploitation workflows with limited human direction, generating “thousands of executed commands across dozens of sessions.” It also cites an industry report on a breach involving nine Mexican government agencies, describing the use of two commercial AI tools, Claude Code and GPT-4.1, to carry out and support attack activity across 34 sessions, generating 5,317 AI-executed commands.
The report also claims the vulnerability response window is shrinking, stating AI can turn a newly disclosed vulnerability into a working exploit within hours. It adds that some government authorities have shortened mandated remediation timelines to as little as 12 hours for critical internet-facing systems.
Prompt injection is flagged as a growing enterprise risk as AI becomes an attack surface in its own right. Check Point Research said detections of “long, malicious prompt-injection payloads” rose roughly fivefold between March and May 2026, which it described as consistent with indirect prompt injection becoming a routine attack path rather than a theoretical concern.
Digital identity is also highlighted as a weakening control in a deepfake-driven environment. The report states that voice, face, documents and real-time video can be convincingly synthesised, and that highly trained reviewers correctly detected about 41% of AI-generated faces. It suggests organisations will need to move beyond visual verification and rely on stronger identity assurance methods such as MFA and out-of-band verification.
On enterprise adoption, the report claims “high-risk” AI prompts doubled over the year, from about one in every 50 interactions to one in every 25. It also says the average organisation runs 10 AI applications a month, many without formal approval, and that between 87% and 93% experience at least one high-risk AI interaction monthly.
Another theme is data exposure tied to everyday AI usage. The report argues that most enterprise data exposure linked to AI comes from approved, routine use rather than direct attacks, as staff “share more context than they realize” to get useful responses.
Lotem Finkelstein, vice president of Check Point Research, said: “A year ago we described AI as a force multiplier for attackers. What we documented this year is more significant: AI has crossed into the live attack chain and is now running operations as a sole operation, that once required a skilled team. The expertise barrier that separated capable attackers from the rest is disappearing, and defenders can no longer assume a human is setting the pace on the other side. The organisations that stay ahead will be the ones that govern how AI is used, secure the AI systems they now depend on, and defend at machine speed rather than human speed.”
The report sets out three broad response priorities for defenders: protecting AI systems organisations depend on, matching the speed of AI-enabled attacks, and governing workforce use of AI tools to reduce data leakage and unmanaged exposure.
You can read the full report here.
