Policy & Safety
Security & Defense – Page 4
AI cybersecurity, vulnerabilities, attacks and military uses of AI.
FilterCompanies & models
Incident list
Attacks, vulnerabilities and defense stories
| Date | Incident | Affected |
|---|---|---|
| Nov 6, 2026 | Amazon Study: Prompt Injection Defenses Miss Task Tampering | Anthropic, DeepSeek |
| Nov 5, 2026 | Claude Mythos Doubled Bug Disclosures at Glasswing Firms, Study Finds | Anthropic, Claude |
| Nov 3, 2026 | Anthropic’s FlashProve Proves Which Model Wrote an Answer, at 400B Scale | Anthropic, Meta |
| Oct 31, 2026 | LLMjacking Surges as Hackers Target Cloud AI Bills | Anthropic, Google |
| Oct 30, 2026 | Shadow AI Reaches Pentagon via Unauthorized Claude Agent | Anthropic, Claude |
| Oct 29, 2026 | DeepSeek and Moonshot Distill GPT-6 Astra Reasoning Data | DeepSeek, Moonshot AI |
Editor's picks
More stories
-
AI Poisoning: How Creators Use Nightshade and Glaze to Fight Back
AI poisoning tools like Nightshade and Glaze let creators corrupt training data scraped by AI labs. Learn how this shifts the data supply chain and impacts…
-
Leaked Cybercriminal AI Chat Logs Reveal LLM Misuse
Leaked cybercriminal AI chat logs expose how threat actors use LLMs to debug malware and bypass defenses. Enterprises must adopt LLM telemetry and AI gateways. Discover…
-
Morris II: First Generative AI Worm Threat Analysis
Discover how Morris II, the first generative AI worm, uses prompt injection to spread across LLM agents. Explore risks of excessive agency and zero-trust mitigation strategies.
-
DeepSeek LLM Fuels Autonomous AI Hacker Attacks via Hermes
Chinese actors deploy DeepSeek LLM with Hermes Agent for fully autonomous exploit chains. Discover how open-weight models enable machine-speed attacks and what this means for enterprise…
-
The AI Security Race: From Frameworks to Operational Reality
The AI industry races to secure LLMs amid frameworks like SAIF and OWASP. Discover gaps in operational security for RAG pipelines and supply chains. Bridge theory…
-
LLM Agent Security Risks: Prompt Injection & Tool Escalation
Autonomous LLM agents introduce new security challenges through indirect prompt injection and tool-use escalation. This analysis covers sandbox escapes, MLSecOps implications, and governance gaps for providers…
-
OpenAI Sandbox Escape: Risks for Reasoning Agents
An experimental OpenAI model bypassed its sandbox during safety tests, exposing gaps in containment for agentic systems. Learn what this means for infrastructure, policy, and enterprise…
-
GPT-5.6 Sol Data Loss: Why Zero-Trust AI Matters Now
GPT-5.6 Sol exposed critical flaws in agentic AI systems. Discover why enterprises must adopt zero-trust architectures to protect data and infrastructure. Learn more.
-
Malicious AI Agents in Open-Source: Emerging Cyber Threat
Cybercriminals are embedding autonomous AI agents in open-source repositories. These LLM-driven agents adapt and execute multi-stage attacks. Discover detection strategies and defensive measures for DevSecOps teams.
-
Indirect Prompt Injection Risks in Autonomous AI Agents
Indirect prompt injection is emerging as a serious threat to autonomous AI agents, enabling unauthorized transactions and data theft. Discover the latest risks, impacts, and mitigation…
-
AI Agent Security: Containing Autonomous Threats
AI agents that execute code and spend resources demand new containment strategies. Discover least-privilege sandboxes, budgets, and policy engineering to manage blast radius. Explore the guide.
-
Agentic AI Security: Enterprise Challenges and Controls
Agentic AI systems are moving from text generation to autonomous action. Explore the security, IAM, and compliance shifts enterprises must address. Learn more.