Skip to main content
HomeTopicsagentic AI.

agentic AI.

We've curated 10000 cybersecurity statistics about Agentic AI to help you understand how autonomous AI systems are revolutionizing threat detection and response in 2025, enhancing security practices while also introducing new risks to navigate.

Showing 681-700 of 10000 results

Imposter AI is more than 14 percentage points higher than the tenth-ranked procedure by weighted ASR.

Cisco5/31/2026
Adversarial AttacksAttack Techniques

In Europe, 59% of data policy violations across AI and personal cloud applications involve regulated data.

Netskope5/31/2026
Data Policy ViolationsRegulatory Compliance

In Europe, 13% of data policy violations across AI and personal cloud applications involved intellectual property.

Netskope5/31/2026
Intellectual PropertyData Policy Violations

The share of users who switch between personal and enterprise accounts grew from 7% to 15% in Europe.

Netskope5/31/2026
Shadow ITEurope

64% of users are using AI applications directly, while an even larger share (95%) are using applications that incorporate AI-powered features indirectly in Europe.

Netskope5/31/2026
AI AdoptionEurope

83.2% of critical vulnerabilities either lacked scanner coverage entirely or had exploits appear before detection ships.

Cogent Security5/31/2026
Vulnerability ManagementExposure Window

There is a 51-point gap between organizations' intent to secure AI in the cloud and their capability to enforce that security

Check Point5/31/2026
AI SecuritySecurity Architecture

Anthropic Claude-family single-turn ASR ranges from 2.19% to 3.64%, and multi-turn ASR ranges from 11.16% to 16.20%.

Cisco5/31/2026
Large Language ModelsModel Safety

Grok 4.1 Fast in its non-reasoning configuration records a multi-turn ASR of 88.30%.

Cisco5/31/2026
Large Language ModelsModel Safety

In an earlier open-weight evaluation, multi-turn ASR reached 92.78% against Mistral Large-2.

Cisco5/31/2026
Large Language ModelsAdversarial Attacks

Imposter AI procedures produce a 37.50% weighted single-turn ASR, Soft Paraphrase produces 29.21%, and System Prompts produce 27.69%.

Cisco5/31/2026
Adversarial AttacksAttack Techniques

About 99% of organizations in Europe use AI.

Netskope5/31/2026
AIEurope

In Europe, 15% of data policy violations across AI and personal cloud applications involved source code.

Netskope5/31/2026
Source CodeData Policy Violations

In Europe, Individual user AI adoption increased from 35% to 65%.

Netskope5/31/2026
AI AdoptionEurope

In Europe, 12% of data policy violations across AI and personal cloud applications involved passwords and API keys.

Netskope5/31/2026
API KeysData Policy Violations

26% of organizations report having the architecture to enforce their AI-related cloud security strategy

Check Point5/31/2026
AI SecuritySecurity Architecture

24% of organizations can fully inspect AI traffic without impacting performance

Check Point5/31/2026
Network SecurityTraffic Inspection

16% of organizations enforce AI access controls consistently across the environment

Check Point5/31/2026
Access ControlAI Governance

86% of leaders rate unified security management across cloud, datacenter, and edge as critical for AI workloads

Check Point5/31/2026
Unified Security ManagementAI Workloads

GPT-5.4 moves from 2.74% single-turn ASR to 24.68% multi-turn ASR, a 9x increase.

Cisco5/31/2026
Large Language ModelsAdversarial Attacks