PPhying中文
Curated security research, vulnerabilities, advisories and tools for practitioners.

Gemini 4: A New Hope

tenzai.com | blog | #llm-security | #review | #gemini | #tenzai
📄 Raw .md⬇️ Download .md🔗 Original ↗

Summary

Google’s newest model just took #1 on Tenzai’s Vulnerability Discovery leaderboard. We got early access and ran it through long, multi-step attacks to see if it holds up.
Published
Collected

Related coverage

blog · tenzai.com
Model Sets: Balancing Cost and Intelligence (Or: Not Every Test Needs Genius)
Tenzai curates and continuously benchmarks sets of frontier and open-weight models that security teams can match to every test type or application risk level, to spend where it matters and save where it doesn't.
blog · tenzai.com
Interactive Testing: Lead a Team of AI Hackers with Tenzai
# Interactive Testing: Lead a Team of AI Hackers with Tenzai Daniel Schwartzman Senior Product Manager Share While Tenzai has the ability to run autonomously, what happens when the human security...
blog · tenzai.com
Tenzai Integrates with Palo Alto Networks to Help Enterprises Validate Which Exposures Are Truly Exploitable
Tenzai's autonomous offensive testing now integrates with Palo Alto Networks Cortex Exposure Management, verifying which exposures are truly exploitable and writing results back into Cortex issues.
blog · tenzai.com
Autonomous WAF Mitigation: From Validated Exploit to Deployed Defense, Fast
Tenzai's autonomous AI hacker now closes the loop from validated exploit to defense, generating and deploying targeted WAF mitigations (including AWS WAF) and guiding permanent fixes at machine speed.
blog · tenzai.com
Incremental Testing: Continuous Pentesting at the Speed of Development
Tenzai explains incremental testing: continuous AI pentesting focused on what each release changed and integrated into CI/CD, catching each vulnerability at the next deployment, not weeks later.
blog · tenzai.com
Introducing Automation Rules: From exploit proof to validated fix, automatically.
Tenzai's Automation Rules close the gap between exploit proof and validated fix: findings route to coding agents with fix prompts, validation tests run after merge, and results post back to tickets.
Back