Jailbroken Frontier Models Retain Their Capabilities

As language model safeguards become more robust, attackers are pushed toward developing increasingly complex jailbreaks. Prior work has found that this complexity imposes a "jailbreak tax" that degrades the target model's task performance. We show that this tax scales inversely with model...

Jailbreaking Frontier Foundation Models through Intention Deception

Large vision-language models exhibit remarkable capability but remain highly susceptible to jailbreaking. Existing safety training approaches aim to have the model learn a refusal boundary between safe and unsafe, based on the user's intent. It has been found that this binary training regime ofte...

Mythos and the Unverified Cage: Z3-Based Pre-Deployment Verification for Frontier-Model Sandbox Infrastructure

The April 2026 Claude Mythos sandbox escape exposed a critical weakness in frontier AI containment: the infrastructure surrounding advanced models remains susceptible to formally characterizable arithmetic vulnerabilities. Anthropic has not publicly characterized the escape vector; some secondary...

Towards Optimal Agentic Architectures for Offensive Security Tasks

Agentic security systems increasingly audit live targets with tool-using LLMs, but prior systems fix a single coordination topology, leaving unclear when additional agents help and when they only add cost. We treat topology choice as an empirical systems question. We introduce a controlled...

System Card: Claude Mythos Preview

This System Card describes Claude Mythos Preview, a large language model from Anthropic. Mythos Preview is their most capable frontier model to date, and shows a striking leap in scores on many evaluation benchmarks compared to their previous frontier model, Claude Opus 4.6. This System Card...

Killer robots are here. Now what? (Lock and Code S07E07)

Big news : Lock and Code is nominated for a Webby Award! You can help us win the People's Voice Award by voting here. Vote now! This week on the Lock and Code podcast … We have to talk about killer robots. No, not the Terminator, and not some Boston Dynamics robot run amok. We have to talk instea...

Perceptual Gaps: ASCII Art and Overlapping Audio As CAPTCHA

As multimodal large language models LLMs advance, traditional CAPTCHAs have become obsolete at distinguishing humans from bots. To address this shift, this paper aims to investigate the possibility of using tasks for which humans have evolved highly specialised neural processing. We introduce two...

PT-2026-28798

OpenClaw before 2026.3.22 contains a policy bypass vulnerability where queued node actions are not revalidated against current command policy when delivered. Attackers can exploit stale allowlists or declarations that survive policy tightening to execute unauthorized commands...

T-MAP: Red-Teaming LLM Agents with Trajectory-Aware Evolutionary Search

While prior red-teaming efforts have focused on eliciting harmful text outputs from large language models LLMs, such approaches fail to capture agent-specific vulnerabilities that emerge through multi-step tool execution, particularly in rapidly growing ecosystems such as the Model Context Protoc...

Secure agentic AI end-to-end

Next week, RSAC™ Conference celebrates its 35-year anniversary as a forum that brings the security community together to address new challenges and embrace opportunities in our quest to make the world a safer place for all. As we look towards that milestone, agentic AI is reshaping industries...

How Vulnerable Are AI Agents to Indirect Prompt Injections? Insights from a Large-Scale Public Competition

LLM based agents are increasingly deployed in high stakes settings where they process external data sources such as emails, documents, and code repositories. This creates exposure to indirect prompt injection attacks, where adversarial instructions embedded in external content manipulate agent...

Security Considerations for Artificial Intelligence Agents

This article, a lightly adapted version of Perplexity's response to NIST/CAISI Request for Information 2025-0035, details our observations and recommendations concerning the security of frontier AI agents. These insights are informed by Perplexity's experience operating general-purpose agentic...

Secure agentic AI for your Frontier Transformation

Today we shared the next step to make Frontier Transformation real for customers across every industry with Wave 3 of Microsoft 365 Copilot, Microsoft Agent 365, and Microsoft 365 E7: The Frontier Suite. Introducing the First Frontier Suite built on Intelligence and Trust As our customers rapidly...

Secure agentic AI for your Frontier Transformation

Today we shared the next step to make Frontier Transformation real for customers across every industry with Wave 3 of Microsoft 365 Copilot, Microsoft Agent 365, and Microsoft 365 E7: The Frontier Suite. Introducing the First Frontier Suite built on Intelligence and Trust As our customers rapidly...

EVMbench: Evaluating AI Agents on Smart Contract Security

Smart contracts on public blockchains now manage large amounts of value, and vulnerabilities in these systems can lead to substantial losses. As AI agents become more capable at reading, writing, and running code, it is natural to ask how well they can already navigate this landscape, both in way...

ZeroDayBench: Evaluating LLM Agents on Unseen Zero-Day Vulnerabilities for Cyberdefense

Large language models LLMs are increasingly being deployed as software engineering agents that autonomously contribute to repositories. A major benefit these agents present is their ability to find and patch security vulnerabilities in the codebases they oversee. To estimate the capability of...

AI Arms and Influence: Frontier Models Exhibit Sophisticated Reasoning in Simulated Nuclear Crises

Today's leading AI models engage in sophisticated behaviour when placed in strategic competition. They spontaneously attempt deception, signaling intentions they do not intend to follow; they demonstrate rich theory of mind, reasoning about adversary beliefs and anticipating their actions; and th...

In-Context Autonomous Network Incident Response: An End-To-End Large Language Model Agent Approach

Rapidly evolving cyberattacks demand incident response systems that can autonomously learn and adapt to changing threats. Prior work has extensively explored the reinforcement learning approach, which involves learning response strategies through extensive simulation of the incident. While this...

Bypassing AI Control Protocols Via Agent-As-A-Proxy Attacks

As AI agents automate critical workloads, they remain vulnerable to indirect prompt injection IPI attacks. Current defenses rely on monitoring protocols that jointly evaluate an agent's Chain-of-Thought CoT and tool-use actions to ensure alignment with user intent. We demonstrate that these...

Jailbreaking LLMs Via Calibration

Safety alignment in Large Language Models LLMs often creates a systematic discrepancy between a model's aligned output and the underlying pre-aligned data distribution. We propose a framework in which the effect of safety alignment on next-token prediction is modeled as a systematic distortion of...