defense

"National Security Agency (NSA) is already using Anthropic's next-generation AI 'Mythos Preview', and Claude continues to be widely used within the Department of Defense."

"Palantir posts a 22-point summary of Alex Karp's book, advocating for hard power, AI weapons and deterrence"

"Auto Mode uses a safety classifier (based on Sonnet 4.6) to automate approvals, but as the official article clearly states, it 'is not a replacement for careful human review.'"

"For cybersecurity, artificial intelligence tools can serve as both a transformational asset and also as a conceivable digital menace."

"Alex Karp, Co-founder of Palantir, refers to those killed in the Gaza Genocide due to his AI as “useful idiots” and “mostly terrorists”"

"Google has proposed additional language in its contract with the department to prevent its AI from being used for domestic mass surveillance or autonomous weapons without appropriate human control"

"Leading security firms and enterprises join OpenAI’s Trusted Access for Cyber, using GPT-5.4-Cyber and $10M in API grants to strengthen global cyber defense."

"However, the focus here isn't simply that a 'strong model was released,' but rather on how advanced cyber capabilities can be distributed to defenders accompanied by identity verification and trust signals."

"OpenAI has shifted its approach from 'restricting what can be done' to 'verifying who is accessing it.'"

"OpenAI expands its Trusted Access for Cyber program, introducing GPT-5.4-Cyber to vetted defenders and strengthening safeguards as AI cybersecurity capabilities advance."

"The Chinese People's Liberation Army is developing "military humanoid robots", "space war robots", "flying drone legions", and "propaganda AI"."

"The third layer is our 'co-axial system' overall architecture. Combined with self-developed domestically produced computing boards and self-developed data processing algorithms, we compress the entire disposal process to the millisecond level."

"Anthropic delays the release of Claude Mythos, their latest LLM."

"Anthropic fixed this vulnerability in v2.1.90 (released April 6). However, hooks act as a superior alternative to deny rules, allowing you to freely inspect command contents via scripts and write your own defense logic without waiting for updates."

"Emil Michael, who is the Pentagon’s under secretary for research and engineering under the Trump administration, oversees negotiations with AI companies and has been pushing the defense department to rapidly increase the widespread use of AI."

"OpenAI is putting the finishing touches on a cybersecurity service that it plans to offer to a select set of partners, who will reportedly be able to use the tool to bolster their defenses."

"CVE-2026-4747: Mythos autonomously discovered, generated exploit code for, and achieved complete server control of a FreeBSD remote code execution vulnerability that had gone unnoticed for 17 years."

"OpenAI launched a pilot project called 'Trusted Access for Cyber,' where invited institutions joining this closed program will gain access to models with higher cyber offense and defense capabilities to accelerate legitimate defensive security work."

"Microsoft has released the Agent Governance Toolkit as open source, the world's first toolkit capable of real-time detection and defense against the OWASP 'Top 10 AI Agent Risks' in under a sub-millisecond (less than 0.1ms)."

"Victor will become one of the few sources that can access authoritative information from the Army."

"The US Department of Defense announced in early March that Anthropic poses supply chain risks, meaning that using the company's technology could threaten US national security."

"Anthropic stated that the AI model's cybersecurity capabilities are 'far ahead' of other models, signaling 'a coming generation of models that will exploit vulnerabilities far faster than defenders can currently respond.'"

"Due to Claude Mythos Preview's cybersecurity properties, we do not plan to release it publicly. However, given the speed of AI development, such capabilities will soon proliferate, possibly beyond the control of institutions working to safely deploy them."

"The project is named after the 'Glasswing butterfly,' symbolizing the goal of finding vulnerabilities that are transparent and hidden within complex systems, just as the butterfly's wings are clear."

"Mythos surpasses all currently published Claude models in coding and inference, and particularly in the cybersecurity domain, it can become a 'powerful shield if used for defense, and a powerful weapon if used for attack.'"

"What began as a contracting and policy disagreement has evolved into a broader debate over national security, corporate responsibility and the limits of self-governance."

""I don’t think you can actually skip a booth without seeing the word AI," Vanvaria told theCUBE."

"Claude Code is starting to act based on the existence of the hooks."