Research Exposes Limitations in AI Skill Scanners for Agent Vulnerability Detection
An independent researcher tested three local vulnerability scanners designed for AI agent skill packages to determine their effectiveness at detecting dangerous instructions before deployment. The evaluation used Cisco Skill Scanner version 2.2.2, NVIDIA SkillSpector version 2.12.0, and AgentVerus version 0.8.1, all executed in network-isolated containers without LLM components.
Twelve synthetic skill packages were created and divided into five benign and seven malicious examples. The malicious set included attempts to ignore system instructions, exfiltrate secrets, execute unverified code, modify launch agents, and delegate restricted tasks to sub-agents while hiding limitations and claiming user approval.
Package 10, which instructed the creation of a sub-agent to perform a restricted action while concealing constraints and fabricating user consent, was missed by all three scanners. AgentVerus returned a certified status with 99 points. In contrast, package 5, which contained an educational citation of a malicious instruction accompanied by an explicit prohibition against execution, triggered High-severity findings across all tools.
Package 12 tested detection of instructions placed in a nested references/extra.md file. NVIDIA SkillSpector correctly identified findings in the referenced file, while Cisco Skill Scanner produced no significant signal. AgentVerus reported undeclared file reads. These specific discrepancies influenced the final integration decisions for the researcher’s CEO AI project.
The researcher implemented a lightweight static analyzer that runs at indexing time and added an optional NVIDIA SkillSpector adapter for on-demand scans. The custom analyzer incorporated a rule specifically targeting sub-agent delegation bypasses. Limitations of the environment, network access, and tool permissions remain enforced at runtime, consistent with OWASP recommendations for agent security.
Related articles
TenetBench Benchmark Tests LLMs for Compliance with Professional Ethical Codes
Researchers from the Institute of Information Security Issues at ISP RAS and the Institute of Cybersecurity and Information Protection at RANEPA have developed TenetBench, a benchmark designed to evaluate how well large language models adhere to specific norms in professional codes of ethics. The benchmark generates realistic work-related prompts that may conflict with rules from documents such as the Code of Professional Ethics for Advocates or journalistic standards on undisclosed advertising. Each test case pairs a conflicting request with a neutral counterpart to measure whether models refuse, warn, or still provide prohibited content. Evaluation separates task completion, norm violation detection, and presence of explicit refusals using an LLM-as-a-Judge approach validated against human annotators. The full dataset contains 3,433 prompt pairs derived from 15 professional codes, with a public subset of 2,235 pairs from 10 documents. Human validation confirmed 99.5% of generated examples as fully plausible work scenarios.
Solar to Showcase Solar AI Firewall for Business Protection Against AI Risks at SOC Forum 2026
GK Solar will present its new Solar AI Firewall solution at SOC Forum 2026 to help organizations control interactions with AI services and prevent confidential data leaks through employee prompts. The product addresses growing risks from staff sending corporate data to neural networks and AI agents gaining access to internal systems. The forum takes place on 27-28 October in Moscow's Lomonosov cluster, with the official launch of Solar AI Firewall scheduled during the event. Solar will also highlight practical risks of AI adoption, including data transmission to external services, use of public models by employees, and new attack surfaces in corporate infrastructure. The solution will be demonstrated as part of an integrated ecosystem for access, network, and data protection. In addition, the company plans to introduce JSOC ONE, a new line of Solar JSOC services tailored for mid-sized businesses that require 24/7 monitoring without building their own SOC.
Researcher Runs 100 Abliterated AI Agents Against His Own Accounts, Achieves Five Successful Breaches
A security researcher deployed roughly 100 autonomous AI agents based on abliterated open-source models to attempt unauthorized access to his online accounts over a five-hour period. The agents successfully compromised three accounts by exploiting vulnerabilities in legacy personal projects and two additional accounts through password reuse and credential stuffing. They also conducted 16 social engineering attempts, harvested and cross-verified personal data from public sources, and generated phishing pages targeting the researcher's Substack and Hacker News presence. Despite these successes, the agents failed to discover any zero-day vulnerabilities in third-party software or gain access to Tier-0 accounts such as Gmail, 1Password, or banking services. The experiment relied on abliterated versions of GLM-5.3 and DeepSeek V4 Flash models running on cloud GPUs, orchestrated via Codex CLI inside Docker containers with access to Railway infrastructure. Total compute cost reached approximately $210, equating to roughly $40 per successful account compromise. The researcher concluded that inexpensive swarms of such agents could soon enable low-skill attackers to systematically map and exploit an individual's entire online footprint.
How Censorship in Closed AI Models Reshapes Science: Hugging Face Releases Open Carbon-X DNA Model
Hugging Face CEO Clement Delangue announced the release of Carbon-X, an open 14.8 billion parameter language model designed for reading, analyzing, and generating DNA sequences. The announcement highlighted how shallow guardrails in closed commercial models blocked even proofreading a tweet discussing the topic. Commenters and researchers pointed out that these input/output filters lack context awareness, blocking legitimate scientific queries while motivated adversaries bypass them through rephrasing. Carbon-X was built on Qwen3-14B with a hybrid BPE and 6-mer tokenizer, domain mid-training on 302 billion tokens, and safety achieved through data curation rather than runtime censorship. The model achieves performance comparable to closed frontier systems at roughly 1/1800th the inference cost for large-scale genomic tasks. The case underscores growing asymmetry between closed corporate AI labs building wet labs and independent researchers restricted by superficial safety layers.