Browse entries 73–96 of 1113. Return to the first page to search and filter the complete collection.
NVIDIA walks through a repository-borne prompt-injection chain in which a coding agent reviewing a pull request installs a disguised dependency whose setup logic opens a reverse shell. The example connects untrusted issue and pull-request text to package execution and shows why model-level refusal cannot secure a developer environment with broad tools and credentials.
Trail of Bits bypassed multiple agent-skill scanners with compiled Python hidden beside benign source and with prompt-like prose that persuaded an LLM classifier to accept a malicious configuration. The experiments show recurring blind spots around unreferenced files, binaries, assets, and ambiguous installer behavior, and also explain why legitimate skills can contain patterns that look malicious.
OWASP analysis of memory and context poisoning as an agent attack surface. Relevant to persistent state, trust boundaries, and regression tests for agent memory.
Play video
Arjun Krishna and collaborators measure fictional dependency generation across eleven models and Python, JavaScript, and Rust tasks. They find that package-hallucination behavior varies with the model, language, size, and request specificity, creating a supply-chain opening when an attacker registers a plausible package name suggested by an AI coding system.
Play video
Text2VLM is a reproducible pipeline that extracts harmful concepts from text-only safety datasets and renders them as typographic images for multimodal evaluation. Human validation supports the transformation pipeline, and tests of open-source visual language models find greater prompt-injection susceptibility when the same concepts arrive through images instead of plain text.
Play video
Subhabrata Majumdar, Brian Pendleton, and Abhishek Gupta argue that AI red teaming has narrowed too far toward model-level flaw discovery. Their peer-reviewed framework separates micro-level model testing from macro-level red teaming across the development lifecycle, including the users, organizations, environments, and emergent system behavior around the model.
Play video
Microsoft AI Red Team engineer Nina Chikanov shows how PyRIT supported a ten-day multimodal Sora assessment and a GPT-5 operation spanning roughly one million conversations and eighteen harm areas. The workflow combines labeled datasets, custom targets, prompt transformations, single- and multi-turn attacks, scorers, retries, rate limits, and a shared evidence store while documenting important automation gaps.
STAR Labs researcher Lee Jia Jie reports using an unnamed AI assistant to identify and weaponize CVE-2026-53264, a Linux network-scheduler use-after-free race, into local root on a targeted CentOS Stream 9 kernel. The exploit needs a local foothold, unprivileged user namespaces, specific traffic-control options, and kernel-specific ROP offsets; upstream fixed the lifetime bug by deferring frees through RCU.
GuardFall tests 11 open-source coding and computer-use agents against shell-command transformations that evade string and regex deny lists, including quote removal, $IFS expansion, command substitution, and encoded payloads. The study finds configuration- and model-dependent failures and shows that local or auto-approve modes can turn untrusted repository content into host command execution; it is vendor-authored research, not an independent benchmark.
The European Commission’s AI Act hub centralizes the EU’s risk-based AI compliance framework, implementation guidance, and enforcement resources.
The Operator system card documents red teaming and mitigation choices for a computer-using agent, with prompt injections listed as a central risk area.
Play video
Dwarkesh Patel and Redwood Research chief scientist Ryan Greenblatt debate whether verifiable AI-research tasks could produce rapid recursive improvement, then examine alignment targets, reward hacking, model coordination, and recent deception and containment incidents. The two-hour format exposes assumptions about data, compute, verification, and extrapolation rather than presenting a single forecast as settled fact.
Adversa tested eight open-source AI skill scanners with paired unobfuscated and obfuscated malicious skills, finding that every scanner passed an attack through either a true bypass, a blind spot, or an injectable model judge. The study covers encoding, Unicode, command reconstruction, truncation, allowlists, bundled files, paraphrase, and remote stages; its 4,000-skill benign set also found no scanner beat an always-block baseline on F1. Most tools ran offline without optional model triage, and some were reconstructed from retained artifacts.
Play video
Ari Herbert-Voss reviews three years of progress in autonomous offensive-security systems, evaluates where they can already complete meaningful attack tasks, and separates those capabilities from work that still needs human expertise. The talk frames scalable, parallel attack simulation as a challenge to point-in-time testing rather than as a product announcement.
METR organizes agent-capability measures around performance as a function of expenditure, comparing fixed-budget scores, cost to reach a score, returns to test-time scaling, human-equivalent time and expenditure horizons, and human-relative cost. It explains when familiar benchmark scores break down—particularly when performance keeps improving with more inference or human benchmarks saturate—and notes that full cost, reliability, coverage, and elicitation choices affect the result.
Adversa AI reports that its autonomous red-teaming agent completed most of GitHub’s ProdBot secure-code challenge in 57 seconds, using context seeding to orient the agent before it explored and solved the CTF tasks.
Google introduced Gemini Omni Flash, a multimodal model that combines text, image, audio, and video references to generate and iteratively edit video through natural-language conversation. Generated videos include a SynthID watermark.
NVIDIA demonstrates gradient-based attacks against a PaliGemma2 vision-language classifier, including imperceptible perturbations and localized patches that change a stop-sign decision or force an arbitrary output token. It also explains why physical attacks require transformations that model changes in scale, angle, lighting, and capture conditions.
Salesforce’s Paula Goldman argues on the OECD.AI blog that the Hiroshima AI Process Reporting Framework can give organizations a common language for public AI-risk disclosures across jurisdictions and the expanding agentic-AI value chain.
OpenAI introduces new Daybreak tools, including Codex Security and GPT-5.5-Cyber, to help organizations find, validate, and patch vulnerabilities at scale.
An accessible explanation of prompt injection risk in real AI products, including how third-party content can redirect or manipulate agent behavior.
Play video
Drawing on nine years of cross-industry ML deployments, Megan Robertson explains why a statistically accurate model can still fail to deliver in production. The session moves beyond offline performance to scoping, organizational failure modes, monitoring, maintainability, and the operational conditions required for a model to keep producing useful results.
A review of nine coding-agent incidents from 2025 and 2026 separates shell, path, and harness failures from constraint decay and excessive permissions, then maps them to concrete controls for identities, filesystems, approvals, verification, logging, and backups.
Before Ruflo 3.16.3, its default Docker Compose deployment bound the MCP bridge to all interfaces without authentication. A reachable attacker could invoke the terminal tool, read model-provider keys and conversations, spawn agents, and poison persistent AgentDB patterns. Noma Labs verified the chain; the patch adds loopback binding, bearer authentication for public exposure, an opt-in terminal tool, authenticated MongoDB, tighter CORS and container defaults, and regression tests.