AI Accelerates DevSecOps but Expands Attack Surfaces Across Code, Supply Chains, and Runtime Environments
Security experts warn that AI is dramatically accelerating both software development and the pace of attacks, forcing DevSecOps teams to adopt new verification strategies across the entire software lifecycle.
According to data from Solar, AI-assisted attackers have reduced the window for exploiting vulnerabilities from 63 days in 2019 to only a few hours in 2025. Anton Prokofiev, director of the secure software development control center at Solar, links this acceleration to two additional factors: heavy reliance on third-party libraries and the rise of vibe coding. Attackers now target supply chains rather than individual applications, since compromising one widely used library can affect up to 10,000 organizations.
Dmitry Evdokimov of Luntry notes that AI models behave differently from human developers. Instead of reusing existing libraries, models frequently generate unique code that has never been tested or reviewed by security teams. This creates fresh risk sets because the resulting code lacks coverage from traditional testing phases.
Another concern arises when developers treat AI suggestions as trusted output. Models can produce syntactically correct code that contains outdated cryptography or missing input validation. If these recommendations are accepted without further review, vulnerabilities may persist undetected.
Open-source projects already contain far more vulnerabilities than human reviewers can address. While AI is being applied to vulnerability discovery, experts caution that libraries may carry false “checked” marks when part of the analysis was performed by models of unknown quality.
The volume of generated code, dependencies, and scan results is growing rapidly. Without corresponding acceleration in analysis, backlogs expand faster than security teams can process them. In addition, 41% of confidential information leaked into large language models now consists of source code.
AI is already used for requirements preparation, task distribution, code and test generation, triage of findings, automated code fixes, and dependency management. The Solar appScreener AI plugin, trained on seven years of data from more than 1,000 companies, delivers over 90% accuracy in triage and up to 85% accuracy in preparing fixes, increasing AppSec team capacity tenfold.
Nevertheless, experts stress that critical vulnerabilities and proposed fixes must still be rechecked by humans. Dmitry Chastukhin of Hexway states that LLM-generated code is inherently untrusted and requires verification, ultimately increasing the workload for security specialists.
With the emergence of AI agents, static checks performed before deployment are no longer sufficient. Decisions are made at runtime under the combined influence of the model, prompt, external data, memory, available tools, and infrastructure permissions. New attack surfaces include prompt injection, malicious automation scenarios, and sandbox escapes.
The Hexway ASOC platform aggregates results from static, dynamic, and composition analysis, normalizes findings, removes duplicates, and prioritizes issues using application criticality and exploitability data. Luntry focuses on container and Kubernetes runtime security, enabling prioritization of AI-generated code vulnerabilities and detection of unknown threats during execution.
Experts predict that further AI adoption will bring both more sophisticated attacks and more advanced defensive techniques, urging organizations to address these challenges early rather than attempting to catch up later.
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.