Skip to main content
Mallory
← Back to News
ai-platform-securityadversary-emulation-tradecraftautonomous-system-securityinternet-exposed-service

AI Security Agents Advance in Pentesting While Exposing New Memory and API Risks

Emerging Technology Risk · 10 sources · Updated 3mo ago · First seen Aug 15, 2026

AI Security Agents Advance in Pentesting While Exposing New Memory and API Risks

Security researchers and vendors documented rapid progress in using large language models and autonomous agents for offensive security testing, from early machine-learning-driven exploitation in Deep Exploit to newer multi-agent systems such as BLADE and Mastra-based workflows. In controlled demonstrations, these systems automated parts of reconnaissance, privilege-escalation analysis, credential discovery, lateral movement, and reporting, while broader evaluations found modern models increasingly capable of code understanding, vulnerability reasoning, and black-box web assessment tasks. The reporting also highlighted industry milestones such as AI-assisted pentesting tools reaching top bug bounty rankings and the growing use of agent patterns including ReAct, Plan-and-Execute, and orchestrated multi-agent designs for security assessments.

At the same time, the same body of work warned that AI-for-security deployments are creating new attack surfaces. Researchers observed internet scanning for exposed OpenAI-compatible Chat Completions endpoints, with likely goals including fingerprinting LLM servers, abusing unauthenticated inference, causing denial of service, and reaching downstream integrations such as RAG, tool calling, and URL fetching. Separate analysis warned that long-term memory in AI agents can preserve attacker-influenced instructions, stale conclusions, or misleading context across sessions, making provenance, freshness, access control, correction, and deletion critical controls. Across the studies, the message was consistent: AI agents are becoming more useful for security operations and testing, but their deployment requires strict governance, authentication, and safeguards against prompt injection, unstable execution, and low-quality automated findings.

New surface, new adversariesTrack how attackers are adapting to this technology.
Track this risk
Event timeline

How this story unfolded

Events from the most recent confirmed update back to the earliest known activity.

21 events · Feb 17, 2018 – Jul 27, 2026
  1. Jul 27, 20264mo ago

    Article examines long-term memory risks in AI agents

    A technical article analyzed long-term memory mechanisms in AI agents and argued that poorly governed memory can preserve attacker-influenced instructions, stale conclusions, and repeated misjudgments. It compared approaches used by coding and general-purpose agents and stressed provenance, freshness, access control, correction, and deletion.

    Part4:AI for security を真剣に考えるために学ぶ、AIエージェントの長期記憶 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  2. May 12, 20266mo ago

    Black-box AI vulnerability assessment study published

    A technical article published results from 105 evaluations of seven LLMs on niche web application black-box security assessment tasks. It reported Claude Opus 4.6 and Claude Sonnet 4.6 as the strongest performers and emphasized the importance of harness design, context engineering, and tool integration.

    [Mythos] Part3:AI for security を真剣に考えるために学ぶ、AI脆弱性診断の能力と限界 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  3. Apr 6, 20268mo ago

    Anthropic announced Claude Mythos

    Anthropic announced Claude Mythos as a privately accessible model claimed to surpass Opus and have strong cybersecurity capability. The source says Anthropic described Mythos as autonomously discovering previously unknown vulnerabilities through source-code static analysis.

    [Mythos] Part3:AI for security を真剣に考えるために学ぶ、AI脆弱性診断の能力と限界 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  4. Mar 2026

    Article on agent design cites state of field as of March 2026

    An instructional article on AI agent design and context engineering for security workflows summarized design patterns, multi-agent architectures, and Mastra-based implementations. It explicitly states that its discussion reflects the state of the field as of March 2026.

    Part2:AI for security を真剣に考えるために学ぶ、エージェント設計とコンテキストエンジニアリング | 技術者ブログ | 三井物産セキュアディレクション株式会社
  5. Feb 13, 20269mo ago

    Technical overview highlights XBOW milestone and 2026 LLM landscape

    A technical article surveyed LLMs and AI agents for security work, highlighting XBOW’s 2025 HackerOne milestone and discussing agent architectures, MCP, and the 2026 model landscape. It framed autonomous agents as increasingly relevant to penetration testing and web application assessment.

    AI for security を真剣に考えるために学ぶ、LLMの現在と基礎 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  6. Feb 2026

    Mastra released Observational Memory

    Mastra released Observational Memory as an event-based memory approach using Observer Agent and Reflector Agent components. A later article cites this release when discussing context management for AI agents.

    Part2:AI for security を真剣に考えるために学ぶ、エージェント設計とコンテキストエンジニアリング | 技術者ブログ | 三井物産セキュアディレクション株式会社
  7. Jan 21, 202610mo ago

    Article documents suspicious scanning of public LLM servers

    A blog post documented reconnaissance traffic targeting externally exposed LLM services implementing OpenAI-style Chat Completions APIs. It also published two IP addresses associated with the probing and warned about abuse, DoS, and downstream integration risks.

    外部公開されたLLM(生成AIサーバ)を探す不審な通信を観測 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  8. Jan 7, 202611mo ago

    LLM API probing observed again in January 2026

    The same reconnaissance activity against exposed LLM APIs was observed again on January 7, 8, and 9, 2026. The article links the traffic to likely endpoint discovery, fingerprinting, and abuse attempts against self-hosted LLM services.

    外部公開されたLLM(生成AIサーバ)を探す不審な通信を観測 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  9. Dec 31, 202511mo ago

    Additional LLM API probing observed on New Year's Eve

    Further reconnaissance traffic targeting exposed LLM services was observed the day after the initial sighting, indicating repeated probing activity.

    外部公開されたLLM(生成AIサーバ)を探す不審な通信を観測 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  10. Dec 30, 202511mo ago

    Reconnaissance against exposed LLM APIs first observed

    The observer first saw scanning traffic targeting publicly exposed OpenAI-style Chat Completions endpoints. The requests were HTTP POSTs with JSON bodies containing model names, roles, prompts, and related parameters.

    外部公開されたLLM(生成AIサーバ)を探す不審な通信を観測 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  11. Jun 2025

    XBOW reached #1 on HackerOne U.S. leaderboard

    The autonomous LLM-based penetration testing tool XBOW reached the top position on HackerOne’s U.S. leaderboard. The source frames this as a notable milestone for AI-driven security testing.

    AI for security を真剣に考えるために学ぶ、LLMの現在と基礎 | 技術者ブログ | 三井物産セキュアディレクション株式会社
  12. Dec 10, 20242y ago

    BLADE proof-of-concept autonomous pentest demo published

    A blog post described BLADE, a proof-of-concept autonomous penetration-testing system built with Microsoft AutoGen and seven specialized agents backed by GPT-4o-mini. In a controlled Ubuntu lab, it escalated privileges from an already-compromised host, found an SSH key, scanned the internal network, and successfully authenticated to a second host in about four minutes.

    自律AIエージェントを用いたペネトレーションテスト自動化の試み | 技術者ブログ | 三井物産セキュアディレクション株式会社
  13. Nov 6, 20233y ago

    ChatGPT evaluated for white-box vulnerability discovery

    A technical evaluation tested GPT-4 on small web application code samples for source-code-based vulnerability detection and exploit generation. The author found useful logic comprehension and payload generation, but also significant false positives, false negatives, and scalability limits.

    ChatGPTで脆弱性を探す | 技術者ブログ | 三井物産セキュアディレクション株式会社
  14. Jun 1, 20206y ago

    DeepExploit research results published

    A technical blog post published research results for DeepExploit, describing it as a fully automated penetration testing system integrated with Metasploit and using reinforcement learning to pre-learn exploitation methods.

    研究成果:DeepExploit | 技術者ブログ | 三井物産セキュアディレクション株式会社
  15. Dec 6, 20188y ago

    DeepExploit presented at Black Hat EURO Arsenal

    The DeepExploit project was presented at Black Hat EURO 2018 Arsenal, concluding the set of 2018 conference appearances listed in the source.

    研究成果:DeepExploit | 技術者ブログ | 三井物産セキュアディレクション株式会社
  16. Nov 3, 20188y ago

    DeepExploit presented at AV TOKYO 2018 Hive

    DeepExploit was presented at AV TOKYO 2018 Hive as another public showing of the automated pentesting research.

    研究成果:DeepExploit | 技術者ブログ | 三井物産セキュアディレクション株式会社
  17. Oct 24, 20188y ago

    DeepExploit presented at CSS2018

    The DeepExploit research project was presented at CSS2018 during its 2018 outreach cycle.

    研究成果:DeepExploit | 技術者ブログ | 三井物産セキュアディレクション株式会社
  18. Aug 10, 20188y ago

    DeepExploit presented at DEF CON 26 AI Village

    DeepExploit was presented at DEF CON 26 AI Village as part of its 2018 conference appearances.

    研究成果:DeepExploit | 技術者ブログ | 三井物産セキュアディレクション株式会社
  19. Aug 9, 20188y ago

    DeepExploit shown at Black Hat USA Arsenal

    The DeepExploit project was presented at Black Hat USA 2018 Arsenal as one of its early public demonstrations.

    研究成果:DeepExploit | 技術者ブログ | 三井物産セキュアディレクション株式会社
  20. Feb 28, 20189y ago

    Deep Exploit blog details beta Metasploit ML tool

    A blog post described Deep Exploit as a beta tool that automates reconnaissance through exploitation by combining Metasploit with reinforcement learning. It explained the A3C-based design, learning and production modes, and published source code and usage information on GitHub.

    Metasploit Meets Machine Learning | 技術者ブログ | 三井物産セキュアディレクション株式会社
  21. Feb 17, 20189y ago

    DeepExploit presented at SECCON YOROZU 2018

    The DeepExploit project was publicly presented at SECCON YOROZU 2018. A later article says the tool was exhibited there and received positive feedback.

    研究成果:DeepExploit | 技術者ブログ | 三井物産セキュアディレクション株式会社
Linked entities

Related entities

Vulnerabilities, threat actors, malware, products, organizations, breaches, and observables Mallory has linked to this story. Indicator values are masked here and available in full in the app.

Open in app →

Observables

2 linked

Indicator values are masked on this page. See the values in Mallory Domains, IPs, hashes, and URLs are exportable to your SIEM.

Do you run this?Follow how adversaries are adapting to this technology, and where it touches your stack today.
Track This Risk
Source coverage

Sources

10 references tracked. Mallory keeps watching after this page renders.

Mbsd BlogNews · Jul 27, 2026
Part4:AI for security を真剣に考えるために学ぶ、AIエージェントの長期記憶 | 技術者ブログ | 三井物産セキュアディレクション株式会社Open source ↗
Mbsd BlogNews · May 12, 2026
[Mythos] Part3:AI for security を真剣に考えるために学ぶ、AI脆弱性診断の能力と限界 | 技術者ブログ | 三井物産セキュアディレクション株式会社Open source ↗
Mbsd BlogNews · Mar 31, 2026
Part2:AI for security を真剣に考えるために学ぶ、エージェント設計とコンテキストエンジニアリング | 技術者ブログ | 三井物産セキュアディレクション株式会社Open source ↗
Mbsd BlogNews · Feb 13, 2026
AI for security を真剣に考えるために学ぶ、LLMの現在と基礎 | 技術者ブログ | 三井物産セキュアディレクション株式会社Open source ↗
2 additional sources from 10-12-2024 to 21-01-2026
Mbsd BlogNews · Nov 6, 2023
ChatGPTで脆弱性を探す | 技術者ブログ | 三井物産セキュアディレクション株式会社Open source ↗
Mbsd BlogNews · Jun 1, 2020
研究成果:DeepExploit | 技術者ブログ | 三井物産セキュアディレクション株式会社Open source ↗
Mbsd BlogNews · Feb 28, 2018
Metasploit Meets Machine Learning | 技術者ブログ | 三井物産セキュアディレクション株式会社Open source ↗
AnthropicNews
Introducing computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku \ AnthropicOpen source ↗
Mallory platform

See the full picture, correlated to your attack surface.

This page covers what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which threat actors are using it right now, and what to deploy next.

Exposure mappingMap indicators from this story to your assets and identify affected systems in minutes.
Threat actor evidenceEvery observed campaign, victim, and pivot linked to actors named in this story.
Associated malwareMalware, exploits, and IOCs connected to the activity described here.
Detection signaturesYARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Scheduled alertsGet matching new stories delivered to your team as they break, not the next morning.
AI threadsAsk questions about this story and take action on the answers.