AI model

Llama Guard 3

2 mentions in 7 days · 2 in 30 days · 2 total · first seen Sep 10, 2026 · last 4d ago

Timeline2 articles

New AI Attack Hides Malicious Instructions in Normal-Looking Text to Evade Safety Filters

Check Point researchers show crafted prose hides policy-violating instructions that bypass all tested LLM gatekeepers, including GPT-4o mini and Llama Guard 3.

A new prompt-crafting technique embeds malicious payloads inside grammatical, natural-looking text without Base64, invisible Unicode, or obvious encodings, defeating lightweight pre-screening gatekeepers. In testing, all four evaluated gatekeeper models—gpt-4o-mini-2024-07-18, gpt-oss-safeguard:20b, claude-3-haiku-20240307, and llama-guard3:8b—classified the crafted wrappers as safe at a 100% bypass rate across 23 obfuscated prompts. GPT-5 Thinking in high-reasoning mode recovered and acted on the hidden instruction in 17 of 18 tests (~94.4%), often spending over a minute and multiple Python executions. Researchers recommend paraphrasing untrusted input, hardening gatekeeper policies, and applying defense-in-depth controls for agentic deployments.

GBHackersSep 11, 05:53 UTC · 4d agoAI safety & security 2 sources

PuzzleMask: Abusing Plain Prose as a Covert AI Attack Vector

Check Point details PuzzleMask, a plain-prose technique that bypasses LLM gatekeeper policy checks, letting hidden payloads reach target models unreviewed.

Check Point Research describes PuzzleMask, a prompt-crafting technique that hides policy-violating payloads inside plain-English prose wrappers, bypassing quick LLM-based policy checks without emojis, Base64, or invisible formatting. The researchers tested 23 automated prompts against gatekeepers including GPT-4o-mini, GPT-OSS-Safeguard 20b, Claude 3 Haiku, and Llama Guard 3, and all were classified as safe despite policies that flagged the plain versions. When submitted to GPT-5 in thinking-high mode with a Python interpreter, the target model extracted and acted on the payload in over 90% of trials. The technique is not itself a jailbreak but can carry a jailbreak prompt as payload; mitigations include input paraphrasing, hardened gatekeeper policies, and output monitoring.

Check Point Researchupdated Sep 11, 05:53 UTC · 4d agofirst Sep 10, 14:32 UTC · 5d agoAI safety & security 2 sources

Appears with

organization OpenAI 2 product Claude 3 Haiku 2 product Llama Guard 3 2 ai model GPT-5 2 organization Anthropic 2 product GPT-5 2 organization Meta 2 ai model Claude 3 Haiku 2 product GPT-4o-mini 1 product GPT-OSS-Safeguard 20b 1 product gpt-oss-safeguard:20b 1 product Jetpack Scan 1 vendor Anthropic 1 vendor Check Point 1 vendor Check Point Research 1 vendor Meta 1

Entities are extracted by the model from each article. Watching an entity keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.