AttentionSwitch: Joint Learning for LLM Jailbreak Defense
AttentionSwitch jointly optimizes sparse attention head selection and query-conditioned scaling to defend LLMs against jailbreaks without catastrophic forgetting.
25
- Mentions · 7d6
- Mentions · 30d9
- All mentions11
- First seenSep 16, 2026
- Last mentioned5d ago
Timeline
11 articles · newest firstBenchmarking indirect prompt injection with task-tampering attacks
New benchmark tests task-tampering in indirect prompt injection across 90 cases in three tool-use environments, showing higher success on decision rules and actions than objective attacks.
25
Source-planned honeypots for LLM agents lure attackers to shadow
Static source-planned honeypots route confirmed attacker requests to isolated shadow copies, reducing real-target compromises from 26/40 to 5/40 on CVE-Bench.
25
JailBound FOL-Guided Framework Evaluates LLM Jailbreaks
JailBound combines taxonomy-driven prompt generation with FOL-guided dual-branch search, boosting average ASR by 5.1 points across 46 LLMs.
35
MEvince Knowledge-Constrained Framework Identifies Malware ATT&CK Techniques
MEvince uses shared behavioral representations and evidence-aligned retrieval to improve exact-ID F1 by 17.4% on 610 malware reports.
40
Signed Manifests Protect Requirements in LLM Agent Handoffs
Signed HMAC-SHA256 manifests reduce unsafe commits in multi-agent handoffs from 69.4% to 9.0% across 1440 trajectories.
45
LLM Repository Context Construction Impacts Generated Code Security
Functionality-focused context boosts code pass rates but only by 3.7 pp for security; agents miss safeguards in 64-73% of cases.
20
Local Open-Weight LLMs Solve Most Vulnerability Reproduction Tasks
All tested configurations solve most of six easier CyberGym tasks down to 4B size with 27B required for the rest.
15
LLMs erode trust in software, Kyle Kingsbury warns
Engineers may suspect any new project due to faked signals of maturity.
30
Towards Full Pipeline FP8 Reinforcement Learning for LLMs
A paper on Hugging Face's daily papers describes FP8 reinforcement learning challenges and proposes a solution to fix training instability in LLMs.
70
PS5 Linux lead quits: "a bunch of noobs using LLMs" that "they don't understand"
PS5 Linux lead Andy 'TheFlow0' Nguyen quit the project, blaming LLM vibe-coders who reported a hypervisor bug to Sony for a bounty.
8
Appears with
- Simon Willisonorganization · 1
- FRVRorganization · 1
- RPCS3product · 1
- Sonyvendor · 1
- Kyle Kingsburyorganization · 1
- Hugging Faceorganization · 1
- PS5product · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.