Papers for

software security engineers

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.

Large language models assist automated privacy impact assessments

PIA-Bench: Towards Automated Privacy Impact Assessment with Large Language Models

Abstract: Privacy impact assessment (PIA) is a critical instrument for institutions to proactively identify privacy risks and develop mitigation strategies before system deployment. While mandated across regulatory and institutional contexts, executing PIA requires extensive privacy and technical expertise, posing a particular challenge for teams without access to such resources. Prior work shows the potential of leveraging large language models (LLMs) to assist practitioners' privacy decisions, but little is known about how accurately and reliably LLMs can automate PIA. To this end, we develop PIA-Bench, the first open benchmark for evaluating LLMs on real-world PIAs. We first audited 499 expert-authored PIAs published by US federal agencies and curated 73 structured PIAs, comprising a total of 451 privacy risk and 831 mitigation items, to evaluate LLMs' ability to assess privacy risks and propose mitigations of complex systems. Our results show that off-the-shelf LLMs produce meaningful assessments and identify avenues for future improvement. Finally, we call for improving domain-specific workflows for LLM agents, developing accountable LLM infrastructure, and designing new quality standards for PIAs.

Fri 11 SeptCryptography and Security
The gist
Privacy impact assessments help organizations find and fix privacy problems before launching new systems, but they usually need experts. The authors created a benchmark called PIA-Bench to test how well large language models (LLMs) can do this job by looking at real-world government assessments. They found that current LLMs can produce useful privacy risk identifications and suggest ways to reduce those risks, although there is room for improvement. Their work suggests ways to improve how AI tools are used for privacy checks and calls for better standards and accountability.
Open → 2609.12571v1

Advances in AI agents improve automatic penetration testing success

Big Enough to Break Out: Tracking the Rising Capability of LLM Penetration-Testing Agents

Abstract: Large language model (LLM) agents are increasingly applied to penetration testing, but we still know little about what they can do or how they fail. We compare two PentestGPT-based systems: a legacy human-in-the-loop system running the open-weight Kimi K2.5, and a newer autonomous system running Claude Opus 4.8. Across three public targets, the autonomous system solves all three, including the two the legacy system never finishes. The legacy result is the more surprising of the two. Even on the machines the legacy system fails to solve, it completes about half the subtasks, while running on ordinary university GPUs with no provider guardrails. We can describe the trend but not explain it, since model, harness, autonomy, and memory architecture all change together. Its direction still points to the next question: what will limit these agents as they take on more complex tasks? The usual answer is long-horizon memory, the loss of access to earlier findings during long attack chains. We test it by adding a coverage-memory layer to both systems, and neither improves outcomes. In the legacy stalled runs we could review, the limiting factor appeared to be planning and commitment rather than lost memory: agents held the evidence for a route forward and never turned it into a concrete exploitation hypothesis, which may suggest that offensive capability will advance with agents' ability to plan rather than with better memory. The same subtask scoring that tracks this capability is available to defenders, who can measure it as it rises instead of waiting to meet it in the field.

Wed 9 SeptCryptography and Security
The gist
Testing computer security by finding weaknesses is easier with AI helpers called large language model agents. This study compares an older system using one AI model with human help to a newer, fully automatic system using a different AI. The newer system solves all test challenges, while the older one struggles but still makes partial progress. Surprisingly, the main problem for the AI is not forgetting clues but deciding how to use them to attack. This suggests future improvements in planning will boost AI hacking tools more than memory.
Open → 2609.10780v1