Pillar
State Of Attacks On Genai (2025) - Analyzes real-world attacks on Generative AI applications, revealing critical security vulnerabilities. Key findings indicate a 90% data leakage rate in successful attacks, with adversaries requiring only 42 seconds and an average of 5…
- from
- Annual Security Reports
- added
- 2026-10-10
- likes
- 0
similar
-
Gravitee softwareseni.com
The AI Governance Gap (2026) - Examines the critical governance gap in enterprise artificial intelligence deployments, focusing on the architectural failure to secure autonomous agents with appropriate identity and access controls. Research indicates that 71 percent of large…
-
Indirect Prompt Injection Attacks: Hidden AI Risks crowdstrike.com
CrowdStrike's Dec 2025 analysis of IPI TTPs targeting enterprise GenAI, including attacker-controlled document poisoning, RAG context manipulation, and practical detection signals for SOC workflows.
-
Pentera pentera.io
AI Security Exposure Benchmark (2026) - Investigates the current state of enterprise artificial intelligence adoption and its impact on security posture, highlighting the disconnect between rapid deployment and limited governance. Key findings reveal that while 100 percent of…
-
Synack synack.com
State of Agentic AI in Pentesting (2026) - Examines the integration of agentic artificial intelligence into penetration testing programs to address the scalability limitations of traditional manual security assessments. Key findings reveal that 87 percent of organizations have…
-
Google AI Red Team: Securing AI services.google.com
Google's red team walkthrough of how AI systems are attacked in practice. An early foundational report, useful for framing rather than for current technique.
-
OWASP Top 10 for LLM Applications genai.owasp.org
Risks and mitigations for developing and deploying generative AI applications.
Annual Security Reports › Survey Reports > Industry Trends: “State Of Attacks On Genai (2025) - Analyzes real-world attacks on Generative AI applications, revealing critical security vulnerabilities. Key findings indicate a 90% data leakage rate in successful attacks, with adversaries requiring only 42 seconds and an average of 5…”