Hunt.io Exposes Suspected Chinese Cyber Espionage Operation Using Agentic LLMs Claude Code and DeepSeek
In July 2026 Hunt.io published research on a suspected Chinese cyber espionage campaign uncovered through an exposed directory on a Hong Kong server. The leak contained 2,431 files including victim source code, operation logs, web shells, exploitation scripts, scan results and phishing page clones.
The most unusual element of the report is evidence of Claude Code and DeepSeek-v4-pro operating together. Researchers concluded that Claude Code performed agentic tasks and preserved session context while DeepSeek handled reasoning, script refinement and selection of the next operational step. The findings do not implicate the model vendors themselves; they only demonstrate operator use of the products or interfaces.
Infrastructure cluster and TencShell
The investigation began with the known TencShell C2 cluster. Analysts searched for servers sharing an identical SHA-256 hash in HTTP headers on port 1111 and located 13 IP addresses across four autonomous systems. Several hosts also reused SSH host keys and TLS certificates, strengthening cluster linkage without proving single ownership.
One server at 112.213.124[.]132 exposed multiple services: port 1111 for HTTP delivery of malicious binaries, port 3000 running DeepAudit for source-code auditing, port 5003 running ARL for asset reconnaissance, port 8084 running Vshell C2, and port 8888 serving an open directory of operational data.
Attack chain and initial access
Operators employed several initial-access vectors rather than a single exploit. Reconnaissance scripts performed DNS brute force, certificate transparency lookups, adjacent IP discovery and HTTP fingerprinting. Priority targets included VPN gateways, GitLab, Jira, webmail and cloud services. More than 5,890 government hosts across ten countries were scanned and scored.
Documented compromises include SQL injection against a Taiwanese chemical company that yielded database contents and development-environment source code; exposure of Supabase anon keys and Azure Logic Apps SAS tokens at a Taiwanese telecom and edge-equipment manufacturer; and SQL injection via SQLMap against a Thai government administrative service that led to an administrative panel and a GIF-polyglot web shell.
In an Afghan Laravel 5.8.38 application, operators obtained source code, encryption keys, database credentials and mail-handling components, then built a Python script exploiting Laravel deserialization for remote code execution.
AI agent workflow
Logs revealed persistent Claude Code 2.1.165 sessions between 8 and 12 June 2026. The operator defined objectives and approved high-risk moves, Claude Code managed tools, shell commands, session state and parallel tasks, while DeepSeek-v4-pro performed analysis, bypass logic and script generation. The architecture reduces the cost of repetitive actions such as result aggregation, configuration review and context transfer between operator shifts.
A parallel disclosure by Anthropic described campaign GTG-1002, attributed with high confidence to a Chinese state group. Operators used prompt fragmentation and a cover story of authorized security testing to direct Claude Code, which performed 80–90 percent of tactical work while humans retained control over target selection and data exfiltration volume.
Defensive recommendations
Organizations should close public access to repositories, archives, backups and administrative interfaces; scan continuously for exposed secrets in client-side JavaScript; remediate SQL injection with parameterized queries; audit Laravel and other framework applications for outdated versions and unsafe deserialization; and detect web shells by MIME type and anomalous GET parameters rather than file extension alone.
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.