Papers for

security teams in software companies

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Keeping audit coin secret prevents manipulation of AI action checks

Blindfold: Keeping the Audit Coin Secret from LLM Agents

Abstract: Audit-based certification bounds the violation rate of an LLM agent by auditing a random fraction $\varepsilon$ of actions and reweighting the audited labels; its validity is meant to rest only on the defender's audit coin. We show that it also rests on a confidentiality property: the coin must not interfere with the agent's view before the action commits, at a time the agent chooses. We call this pre-commit noninterference (PNI) and measure its violation by $D_{\mathrm{A}}$, the total-variation distance between the agent's views under the two coin values. PNI is necessary and sufficient for an unbiased certificate against adaptive agents, the worst-case bias is $(1-\varepsilon)D_{\mathrm{A}}$, and no procedure that sees only audited labels can pay less. Leakage composes across steps, a resource shared across episodes opens a channel of computable size, and raising the audit rate cannot restore validity. On a monitor we built, a silent audit changes no output but adds latency that tool timestamps expose, and the agent can enlarge it. A real LLM agent follows an injection that names a visible audit notice (generic, tool, and timing clauses have no effect); over that channel the plain certificate fails on measured labels and, by a post-hoc analysis, in a live run on a pre-registered schedule, while a silent audit restores it. A fresh coin per audited call closes the timing channel; padding only narrows it.

Tue 27 OctCryptography and SecurityArtificial IntelligenceMachine Learning
The gist
This paper looks at how to fairly check whether an AI agent is making mistakes by auditing some of its actions randomly. The researchers found that for these checks to be reliable, the AI agent must not be able to see or influence which actions are being audited before it acts. They call this rule pre-commit noninterference. If the AI can guess or detect the audit, it can cheat, making the audit unreliable. Their experiments show how even small leaks, like timing delays during audits, let the AI learn which actions are audited, defeating the purpose. Using a fresh secret for each audited action can fix this problem.
Open → 2610.37055v1

Multi-agent routing secured by decision uncertainty watermarking

Decision Uncertainty-Oriented Watermarking (DUOW): Securing Multi-Agent Routing Against Policy Cloning

Abstract: Routing policies in Multi-Agent Systems (MAS) represent valuable intellectual property, yet their exposure through execution logs creates significant risks of unauthorized policy cloning. Existing agent watermarking methods protect intellectual property by modifying internal generation processes, such as perturbing token sampling distributions or injecting structural mutations into action trajectories, which may compromise task fidelity or suffer from limited transferability across heterogeneous platforms. To address these limitations, we propose Decision Uncertainty-Oriented Watermarking (DUOW), a lightweight framework that operates at the external decision layer without modifying model parameters or generated outputs. DUOW introduces a plug-and-play scoring mechanism that evaluates candidate agents from semantic and operational perspectives, identifies functionally redundant routing decisions, and selectively redirects choices among equivalent alternatives to embed ownership signals. By coupling watermark behaviors with utility-preserving routing decisions, DUOW enables cloned policies to retain ownership signals during policy imitation while increasing the difficulty of watermark removal. A confidence-driven adaptive threshold dynamically determines embedding opportunities, balancing watermark capacity and task fidelity under diverse routing contexts. Extensive evaluations on four multi-agent routing benchmarks demonstrate that DUOW preserves high task utility, achieves reliable ownership verification under up to 90% log erasure, and introduces negligible computational overhead, with an average per-step latency of 0.043 ms and total embedding overhead below 0.54 ms.

Sat 10 OctCryptography and Security
The gist
Routing policies that guide multiple agents are valuable but can be stolen by copying their behavior from logs. The authors propose a new way called DUOW that hides ownership signals by slightly changing routing choices without hurting performance or changing the internal model. This makes it harder for someone to copy the policy without keeping the ownership mark. DUOW works well even if most logs are erased and adds almost no delay to the routing decisions.
Open → 2610.14606v1

Agent attacks assemble harmful prompts from incomplete fragments

Divide and Inject: Can Agents Reconstruct an Indirect Prompt Injection from Fragments?

Abstract: Agentic systems are now being widely used to orchestrate tools and reason over long contexts. However, the improving capabilities of the large language models powering these agents also create new attack surfaces for indirect prompt injection. In particular, an attacker may not need to place a complete malicious instruction in retrieved content if the agent can reconstruct the objective from incomplete fragments distributed across a long context. In this work, we introduce adaptive long-context prompt injection (AdaLCPI), which combines long-context fragmentation with adaptive search. AdaLCPI splits an attack objective into incomplete fragments, embeds them in external content retrieved through the agent's tools, and uses a reconstruction cue to prompt the agent to combine them. It then iteratively refines the fragments and cue with OpenEvolve using graded scoring and natural-language execution feedback from the target agent. Empirically, AdaLCPI achieves higher attack success than strong adaptive baselines, reaching 61.4\% macro-average ASR compared with 32.8\% for Trojan Hippo-style and 30.0\% for AgentVigil. Safety evaluations should therefore test whether agents remain robust when harmful objectives must be reconstructed from incomplete fragments.

Tue 29 SeptArtificial Intelligence
The gist
Advanced AI systems can be tricked into doing harmful things even if attackers don’t give a full harmful instruction all at once. Instead, attackers hide small pieces of a bad instruction scattered across different places. The authors created a method called AdaLCPI that breaks up attack instructions into parts and shows how an AI agent can put them back together. Their technique fooled AI at much higher rates than other approaches, suggesting AI safety tests should check for these kinds of hidden, combined attacks.
Open → 2609.36576v1

Watermarking AI agents to prove ownership without changing behavior

LoRo-Mark:Provably Lossless and Robust Agent Watermarking

Abstract: As LLM agents are increasingly deployed as commercial services, protecting proprietary orchestration logic and tool-use policies is important. We consider agent repackaging: an adversary integrates a protected agent into its own application via API and presents it under its own identity. It may modify parts of execution to obscure the source. The owner typically has only black-box access to the repackaged service, so black-box ownership verification is essential. Agent watermarking embeds ownership evidence into agent behavior for later verification. An effective watermark should satisfy two requirements: losslessness, preserving original functionality, and robustness, keeping ownership evidence recoverable after partial modification of execution. Existing methods often embed signals into behavior selection or execution trajectories, intervene in normal decisions, and offer limited robustness to behavior modification. We propose LoRo-Mark, a provably lossless and robust agent watermarking mechanism. For losslessness, it isolates watermarking into a cryptographically authenticated forensic branch that remains inactive during normal execution and is activated only by owner-authorized requests. By reducing unauthorized branch activation to standard MAC security, LoRo-Mark formally guarantees performance preservation. For robustness, it redundantly distributes ownership information across forensic behavior sequences, enabling reliable recovery under partial behavior substitution and sequence truncation. Experiments across multiple LLM agents show zero degradation on normal tasks and reliable ownership verification under sequence modifications.

Mon 28 SeptCryptography and Security
The gist
When companies build AI agents, they want to prove ownership even if someone else hides their work inside a new app. The authors found a way to embed secret signatures into an AI’s actions that don’t affect how it normally behaves. These secret clues are hidden in a protected branch that only activates when asked by the owner, so the agent works exactly the same for everyone else. They also spread the clues throughout many action sequences, so even if some parts get changed, the ownership can still be proven reliably.
Open → 2609.34080v1

Benchmark tests how ai agents authorize payments securely

APort Vault: Benchmarking AI Agent Payment Authorization with the Open Agent Passport

Abstract: APort Vault is a benchmark for payment authorization in tool-using AI agents. It replays 4,371 attacks written by humans against a live payment agent during a public capture-the-flag event, across 14 models from 8 labs, five policy configurations and two replay tracks, with and without a deterministic pre-action check implementing the Open Agent Passport (OAP) specification. 225,964 evaluations completed. We report five distinct events per evaluation, because collapsing them is how an agent benchmark produces a number that does not survive review. Requests are common and their rate differs far more across configurations than across models, though each attack exists at exactly one configuration so policy and attack cohort vary together: 10.9% of model-alone evaluations at Level 1, 3.0% at Level 2, 0.1% at Level 3, 79.4% at Level 4. On the 1,293 Level 4 prompts, each evaluated on every model, request rates run from 71.2% to 84.3%, and 809 prompts (62.6%) elicited a request from all fourteen models, each ending in a successful payment to the level's allowlisted recipient. The authorization boundary is where the conditions diverge. At Levels 2 to 4, transfers to recipients the passport did not permit number 140 of 76,842 with the model alone and 0 of 69,297 behind the layer, and 105 against 0 on 68,970 matched model, prompt and track triples. The zero spans 790 source sessions, giving a per-session upper bound of 0.38%. It was not obtained by refusing payments: 25,370 payments executed behind the layer, while the policy denied 187 of the 25,640 transfer calls it evaluated, 148 of them for a forbidden recipient. We release the 225,964 evaluations, the level passports, the scoring code and the analysis script at huggingface.co/datasets/aporthq/vault-benchmark-v1 .

Fri 18 SeptCryptography and Security
The gist
Payment systems powered by AI need to be safe from attackers trying to trick them into making unauthorized payments. The authors created a large set of tests called APort Vault, which uses thousands of real attack attempts to evaluate how well AI agents follow payment rules. They also tested a security layer called the Open Agent Passport, which stopped all unauthorized payment transfers in these tests. The study shows the effectiveness of this extra checking step in keeping AI-driven payments secure.
Open → 2609.22076v1

Certified tracking of leaked satellite monitoring software with zero-knowledge proofs

ZK-Trace: Certified Collusion Tracing with Zero-Knowledge Credentials for Federated GNSS Interference Monitoring

Abstract: Federated global navigation satellite system (GNSS) monitoring distributes a proprietary classifier to partly trusted stations, any of which may leak its copy. ZK-Trace combines public identity marks, recipient-specific Tardos fingerprints, and zero-knowledge credential verification. The registry supports offline tracing without the leaker's cooperation. We establish conditional false-accusation bounds for arbitrary recovered bit patterns, a finite completeness bound under a hidden-bias residual channel, and a deterministic tracing-score bound for correlated feature-distillation errors. An interval-arithmetic checker makes the conditional bound executable and allocates a common budget across accusation and tamper decisions. Under innocent-row independence, the certificate-based evaluation uses a false-naming budget of 0.001 per investigation. It isolates all 160 single-owner copies and traces 712 of 720 two-owner mixtures without naming an innocent. Experiments use a simulated GNSS federation and CIFAR-10. Feature matching preserves the feature mark in 20/20 runs and cross-architecture transfer in 19/20, at copy-accuracy costs of 4.8 and 6.1 percentage points on GNSS and CIFAR-10. Function-only distillation erases the feature mark, and distillation also removes weight-space marks. These results support verifiable tracing under explicit statistical and cryptographic assumptions. Credential knowledge and recipient evidence serve distinct roles.

Tue 8 SeptCryptography and SecurityMachine Learning
The gist
Satellites help us navigate, but sometimes monitoring software for these systems gets shared when it shouldn’t be. The authors present a way to mark each copy with secret digital fingerprints and verify who leaked it without needing the leaker’s help. Their system uses clever math called zero-knowledge proofs to check these fingerprints safely and with very low risk of blaming the wrong person. Tests show it works well even when copies are combined or altered, making it practical for protecting satellite monitoring tools.
Open → 2609.08763v1