PyPI adds security tab for release attestations
Developers can review digital attestations and provenance status for every release file on project pages.
1
Stories per day · 30 days
5 RSS
Relatedagentic-ai2authorization2prompt-injection2watermarking2agent-harness1agentmark1agent-memory1agent-security1ai-agents1alert-validation1apple-m41attestations1automark1benchmark1c2pa1canary1causal-graphs1causal-history1
62 stories, newest first · one card per story
Stories tagged #provenance
Keyed homoglyph substitution for LLM text provenance
KeHoSu encodes attribution in homoglyph variants over copy-paste channels, surviving reordering and partial quotes for accurate platform and account attribution.
Provenance Identity Continuity proves confused-deputy exclusion in causal security graphs
PIC and COSS generalize security-state continuity over causal execution graphs, proving authority cannot be resurrected across branches.
Relational theory of authority separates proof of possession from proof of relationship
The paper introduces Proof of Relationship alongside PoP, using causal histories to classify executors and prove authorization is occurrence-sensitive.
Privacy-preserving language model watermarks with aggregate estimation
Red-Green Sequence watermarks enable accurate population-level frequency estimation while limiting individual privacy leakage.
Launder-Bench benchmark for detecting agentic code laundering
Introduces Launder-Bench with 1,128 pairwise and 229 attribution instances to test provenance under agentic rewriting from 282 repositories.
TextSeal2 watermarks LLM output under speculative decoding without distortion
At equal diversity and a 10^-3 false-positive rate, it detects 8 to 22 points more 100-token answers than TextSeal.
HAIDL models human-AI notebook interactions as durable data
HAIDL-X middleware for JupyterLab stores prompts, responses, feedback, and provenance and derives behavior measures from traces.
Richer situational awareness does not always improve SOC alert validation
On Cadets and Theia, comprehension-level context lifts Recall@5 by as much as 75 points versus Max. CSG.
Partial C2PA credentials used to adapt synthetic-image forensic detectors
Reconstructing the last signer’s edits from partial credentials improved detectors on images postprocessed by C2PA tools.
Black-box signatures proposed to detect LoRA adaptations of diffusion models
Tests use only the intrinsic dimension of generated images and assume unknown adaptations and model families.
SoK finds host intrusion detectors leave time assumptions unstated
A review of detector papers finds most provenance systems state no clock bound and trust stored evidence.
TrFP fingerprints LLMs from how representations change with depth
Fingerprints stay under 0.3 MiB and remain stable after fine-tuning, pruning, quantization, merging, and permutation.
Decode-time provenance binding blocks MCP steganographic prompt injection
On a real MCP testbed, 24 adaptive payloads succeeded 100% undefended and 0% with the proposed mechanism.
SpectralCanary uses low-frequency traces to audit T2I dataset use
With SD3.5 across five datasets it averages 92.4% per-generation TPR at 1% FPR and mean ΔFID of 0.05.
AgentTriage replays security diagnoses across agent runtimes
On 120 cases it reports 0.858 grounded joint diagnosis and a 0.036 unsupported-claim rate.
HiMark adds controllable multi-bit watermarks to LLM text
With 32 bits in 300 tokens, authors report 90.01% extraction accuracy after paraphrasing while preserving text quality.
Context compaction in coding agents can launder forged user instructions
Flattening harnesses accept forged approvals 87-100% of the time, while four role-typed harnesses reject the same attack.
Self-Reflective Chain-of-Thought (SRCoT): Evidence-Based Text Generation via a Closed-Loop Retrieve--Reflect--Revise Pipeline on Edge LLMs
SRCoT is a local retrieve-reflect-revise pipeline for source-labeled LLM outputs, shown only in synthetic case studies.
Optimist Deep Research: An Iterative, State-Centric Agent for Autonomous Literature Research
Optimist Deep Research is an open-source iterative agent that audits literature search through a mutable evidence state.
SAFE-Mem blocks unauthorized use of LLM agent memory
On MPBench and LoCoMo, attack success was at least 77% below the strongest utility-preserving defense.
What Survives Self-Evolution: Provenance Marks in Copied Agent Harnesses
Provenance marks in leaked agent harnesses mostly fail when a self-evolving LLM rewrites the marked code.
WaveMark: Robust Wavelet-Domain Image Watermarking for Diffusion Models
WaveMark watermarks diffusion noise in the wavelet domain and keeps over 88% accuracy under removal attacks.
Two AI-assisted research repos leave incomplete Git provenance
Co-Authored-By marked 66 of 121 AI-related commits in one repo and 1 of 39 agent-note commits in the other.