60
57
30
30
30
30
30
42
45
ReCAST: Restoration-aware Cascaded Stage-wise Training for Obfuscated SMS Risk Classification
ReCAST distills a large teacher model's de-obfuscation ability into smaller models for robust classification of obfuscated Chinese SMS fraud messages.
The paper proposes ReCAST, a restoration-aware cascaded stage-wise training framework for classifying obfuscated Chinese SMS messages. It distills a large teacher model's de-obfuscation capability into a smaller deployable student by supervising obfuscated span detection, obfuscation type prediction, and text restoration, then uses the student for risk classification. On an internally constructed real-world Chinese SMS benchmark, ReCAST substantially outperforms directly trained baselines under obfuscation, targeting production latency and throughput constraints.
28
35
57
35
47
30
30
42
57
57
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.
55
57
30
42
50
42
30
45
30
42
45
55
35
30
42
60
35
30
30
30
30