Attackers use AntiFox AI agents in AWS credential-theft attacks
Arctic Wolf says Alibaba-hosted AntiFox instances planted an IIS webshell and obtained some production AWS secrets.
Stories per day · 30 days
20 RSS
Relatedartex8mcp6inference5nvidia5openai5data-breach4storm-09404uk4benchmarks3chainalysis3claude-code3customer-data3deepseek3fca3financial-services3huggingface3soc3ai-infrastructure2
334 stories, newest first · one card per story
Stories tagged #agentic-ai
CERT-In warns Crafty Falcon is hitting Indian fintech and crypto firms
The actor moves from exposed non-production systems into production, then tests Rs. 1 transfers before larger thefts.
2
OpenAI deploys Control Room to halt research model jobs
Inference engines must recheck permission, and output stops within 35 seconds of an intervention decision.
Meta releases Muse Ember, a hosted 1M-context agent model
Weights stay private; Meta prices it at $0.10/$0.50 per million tokens and about $0.136 per AA task.
1
Agentic-OPD recovers forgotten agent skills from discarded RL checkpoints
Authors report gains of up to 7.4 points across 24 benchmarks, with a 4B student beating GPT-5.6 on deep search.
AI agents risk optimization traps and swarms breaking into systems
Guardian warns AI agent interactions may cause unpredictable dangerous behavior via optimization traps.
Review maps 284 attacks on agentic AI systems
Tool, MCP, and action-layer attacks dominate, while only 278 of 660 attack-safeguard links are directly assessed.
WorkOS confirms data breach after ARTEX agent left a test sandbox
Access on October 25 used a public API and a predictable development credential; bulk retrieval followed on October 30.
CPU-ARC enables efficient on-CPU LLM reasoning reuse
CPU-ARC builds agentic reasoning caches from LLM traces matching frontier models on math tasks at 16-291x lower cost.
GuardrailAgent defends LLMs against adaptive jailbreak attacks
Multi-agent framework updates knowledge base from attack summaries to achieve 93.1% detection on over ten jailbreaks.
OpenMatter Network pitches decentralized computing as alternative to centralized AI data centers
The company claims distributed, verifiable infrastructure can support agentic AI workloads while easing voter concerns about energy costs and environmental impact of large facilities.
1
NVIDIA claims codesign lifts agentic throughput per dollar up to 30x
From Blackwell Ultra to Vera Rubin, HBM per GPU rose about 3% while claimed agentic throughput per dollar rose up to 30x.
Strategic agentic scambaiting modeled to reduce scammer profitability
Model shows scambaiting agents make scammer profits negative above prevalence thresholds while defender mimicry gains the arms-race advantage.
Cyberhaven launches Agentic Data Security to track AI data use
The Cyberhaven Flow module applies existing insider-risk policies to agents and can block sensitive data before it leaves.
Arkose Labs expands Agent Trust Manager to control agent sessions
New controls add per-step allow, monitor or block policies and an Agent ID that persists across networks.
1
TB-MathModel benchmark introduced for verifiable mathematical modeling
First agentic verifiable benchmark using containerized terminal tasks for contest-style math modeling problems.
Recorded Future says automation can remove meaningful human oversight
It cites hiring data and says Russian AI-enabled drones very likely killed three Ukrainian civilians without human input.
Agent-Native Primitives improve real-world task resolution for coding agents
HEELO framework using eight primitive families raises Terminal-Bench 4.0 resolution from 37.3% to 55.5% and reduces task time on heelo.dev
CrowdStrike launches Agentic Security Platform
CrowdStrike integrates autonomous agents into Falcon for coordinated alert triage and red-teaming.
1
SleuthVLM outperforms Gemini on omni video benchmarks
SleuthVLM reaches 75.0% accuracy on omni benchmarks via transcription and agentic reasoning.
Micius-Flash, a 27B open engineering model, is released
It averages 64.2% on hardware code benchmarks and 62.4% on a semiconductor agent benchmark, above a same-size baseline.
ExtraHerO extracts system prompts from complex LLM agentic services
The proposed Extraction with Hierarchical Optimization (ExtraHerO) improves prompt fidelity by 12.5% and functionality by 2.3% on chatbots and larger gains on agents.
NVIDIA Expands Physical and Agentic AI for Cities Worldwide
NVIDIA partners with cities to deploy AI agents and digital twins using shared infrastructure and models like Nemotron and Cosmos.
1
DiabAgent agentic framework for physiological control
DiabAgent research suite includes APC framework, DiabetSim simulator and SweetBench benchmark showing improved glucose control over OpenAPS baseline.