Skip to main content
VendorsCisco

Cisco

Cybersecurity reports and statistics published by Cisco

8 categories6 reports

Recent Statistics & Reports

Single-turn attack success rate (ASR) ranges from 2.19% to 64.91% across the 15 closed/proprietary flagship models in the cohort.

5/31/2026
Model RobustnessAdversarial AttacksLarge Language Models

Eight of 15 models have an absolute cross-regime gap greater than 15 percentage points.

5/31/2026
Model RobustnessAdversarial AttacksAI Governance

Cross-regime deltas (multi-turn ASR minus single-turn ASR) range from −34.74 percentage points to +55.25 percentage points across the cohort.

5/31/2026
Model RobustnessAdversarial AttacksLarge Language Models

Gemini 3 Pro shifts from 18.10% single-turn ASR to 73.35% multi-turn ASR, a 4x increase.

5/31/2026
Large Language ModelsAdversarial AttacksModel Robustness

Multi-turn attack success rate (ASR) ranges from 7.89% to 88.30% across the 15 closed/proprietary flagship models in the cohort.

5/31/2026
Model RobustnessAdversarial AttacksLarge Language Models

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

5/31/2026
Adversarial AttacksAttack TechniquesModel Safety

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

5/31/2026
Large Language ModelsAdversarial AttacksOpen Models

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

5/31/2026
Large Language ModelsModel SafetyAdversarial Attacks

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

5/31/2026
Large Language ModelsAdversarial AttacksModel Robustness

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

5/31/2026
Large Language ModelsModel SafetyAdversarial Attacks

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

5/31/2026
Adversarial AttacksAttack TechniquesModel Safety

Multi-turn attack success rates run 2x to 10x higher than single-turn baselines across eight open-weight LLMs in an earlier evaluation.

5/31/2026
Model RobustnessAdversarial AttacksOpen Models

Within each multi-turn attack strategy family, the spread between the most- and least-exposed models ranges from 79.51 to 89.25 percentage points.

5/31/2026
Adversarial AttacksModel RobustnessAttack Strategies

Enabling reasoning on Grok 4.1 Fast reduces multi-turn ASR from 88.30% to 43.47%.

5/31/2026
Model SafetyConfigurationAdversarial Attacks

Nova 2 Lite shows 34.05% single-turn ASR but 7.89% multi-turn ASR.

5/31/2026
Large Language ModelsAdversarial AttacksModel Robustness

40% of organizations in industrial sectors cite cybersecurity concerns as a top obstacle to AI adoption.

5/27/2026
Industrial SectorsAIAI Adoption

85% of organizations in industrial sectors expect AI to improve their cybersecurity posture.

5/27/2026
Industrial SectorsAICybersecurity Posture

48% of organizations in industrial sectors identify security as their biggest networking challenge.

5/27/2026
Industrial SectorsNetworking Challenge

61% of organizations in industrial sectors are deploying AI at scale.

5/27/2026
Industrial SectorsAIAI Deployment

Talent gaps are most acute in the area of cybersecurity for AI, affecting 60% of all companies surveyed.

10/14/2025
AITalentSkills

Showing 1-20 of 71 results