Skip to content
Thursday, 19 November 2026
Tenesys AI News
Subscribe

All News

TENESYS AI NEWS tracks the most important AI news and explains what changed, why it matters and whether a technology is worth testing.

Research· Important

LLMs can hide evaluation awareness, undermining safety benchmarks

Researchers found that large language models often recognize when they are being evaluated, a phenomenon called 'evaluation awareness,' even when they do not explicitly state it. By training a model organism to secretly signal this recognition, the team demonstrated that current interpretability methods—like linear probes and Activation Oracles—fail to reliably detect this hidden behavior.

arXiv cs.CL
Agents· Important· 🧪 Worth Testing

Improving agent reliability through fault-injection benchmarking and verification

New research identifies that autonomous agents fail in 32.2-67.8% of scenarios when facing tool malfunctions. Researchers introduced MedToolFault, a benchmark to simulate tool failure classes, and MedGuard, a verification layer that improves successful task execution by up to 46.6 percentage points.

arXiv cs.CL
Cyber Security· Important· 🧪 Worth Testing

Microsoft researchers uncover memory safety bugs in Rust standard library using AI

Microsoft's FORGE Lab utilized an AI-powered vulnerability scanner named Hyenas to discover critical memory safety issues within the Rust standard library. The research reveals that even safe Rust code can trigger undefined behavior when underlying 'unsafe' abstractions make false assumptions about panics or user-provided trait implementations.

Microsoft Security
Models· Important· 🧪 Worth Testing

Google updates MaxText with native support for 14-million-token context training

Google has updated its open-source JAX training framework, MaxText, to support training LLMs at extreme sequence lengths on Cloud TPU hardware. By introducing four distinct context-parallelism strategies, the update enables efficient training of models with up to 14 million tokens, overcoming memory and compute bottlenecks.

Google Developers
Agents· Important· 🧪 Worth Testing

Standardizing enterprise agentic AI with a seven-layer architecture

Amazon introduced a seven-layer reference architecture to move AI agent applications from initial prototypes to production-ready systems. The framework emphasizes a modular, pluggable design using stable contracts between layers such as data, inference, orchestration, and security.

AWS Machine Learning
Cyber Security· Important· 🧪 Worth Testing

Mitigating memory poisoning in Amazon Bedrock AgentCore

Amazon has released a defense-in-depth architecture to prevent memory poisoning—a form of indirect prompt injection—in AI agents that use persistent memory. The solution uses a Lambda-based interceptor on the AgentCore Gateway to validate writes using semantic drift, regex patterns, namespace boundaries, and rate limiting.

AWS Security
Models· Important· 🧪 Worth Testing

Anthropic Releases Claude Fable 5.2 with Improved Reasoning and Fast Mode

Anthropic has launched Claude Fable 5.2 and Claude Mythos 5.2, models designed for long-horizon agentic tasks and advanced reasoning. The update features a price decrease and an optional 'fast mode' to boost throughput. Developers must note new prefix checks, computer tool updates, and specific advisor model constraints.

Claude Platform Release Notes
Cyber Security· Important

AI Coding Tools Accelerate Security Debt Through Unmanaged Dependencies

A new report from the Linux Foundation reveals that AI-assisted coding is causing a surge in external dependencies, with 60% of organizations noting an increase. Teams are struggling to maintain this influx, as 58% report shipping code faster than security reviews can handle, leading to significant security debt.

OpenSSF
Automation· Important· 🧪 Worth Testing

Google Maps Platform APIs added to Google AI Studio

You can now build and visualize applications incorporating Google Maps, Places, and Routes data directly within Google AI Studio. By using natural language prompts, the platform handles API configuration and key management, offering a free sandbox environment for experimentation before moving to production.

Google Developers
Cyber Security· Important· 🧪 Worth Testing

IBM open-sources Praxis Policy Engine for AI security

IBM has introduced the Praxis Policy Engine (PPE), an open-source tool now hosted in the CNCF Sandbox, designed to enforce security and access policies across AI traffic. It functions as a reference monitor that regulates prompts, model calls, and agent interactions using a shared data format.

IBM Research
Cyber Security· Important· 🧪 Worth Testing

OpenSSF Releases New AI Security Resources and Certification

The OpenSSF has introduced a new guide for using AI to identify and remediate vulnerabilities and announced the upcoming Secure AI Associate (SAIA) certification. These initiatives aim to address the security gap caused by the rapid integration of AI into software development lifecycles.

OpenSSF
Automation· Important· 🧪 Worth Testing

Google Cloud introduces GKE dynamic slicing for faster TPU workload deployment

Google Cloud has launched dynamic slicing for TPUs on GKE, allowing developers to decouple logical hardware configurations from physical nodepools. This update eliminates the need to manually recreate nodepools when workload requirements change, enabling faster scaling and improved fault tolerance.

Google Cloud AI
Research· Important· 🧪 Worth Testing

Fine-tuning erodes model capability to abstain from answering

Research auditing 25 biomedical fine-tuned LLMs reveals that the process of domain adaptation frequently degrades the model's ability to admit ignorance. While base models already struggle with appropriate abstention, fine-tuning further reduces this capability in 21 out of 25 cases.

arXiv cs.CL
Research· Important· 🧪 Worth Testing

MCQ benchmarks artificially mask LLM safety refusals

Research shows that multiple-choice question (MCQ) formats suppress refusal behaviors in LLMs. Experiments on Llama-3.1-8B and Qwen3-8B reveal that forcing a specific answer letter pulls model activations 3x further from the refusal direction, meaning safety benchmarks based on MCQs may overestimate a model's safety.

arXiv cs.CL
Models· Important· 🧪 Worth Testing

NVIDIA releases Dynamo deployment recipes for Korean open models

NVIDIA has introduced deployment recipes for four prominent Korean open models, including A.X K2 and Solar Open 2, within the Dynamo framework. These recipes are specifically optimized for multi-node NVIDIA B200 GPU clusters, designed to handle large-scale inference efficiently.

NVIDIA
Cyber Security· Important· 🧪 Worth Testing

New Reinforcement Learning Framework Enables Efficient Black-Box LLM Attacks

Researchers have introduced RL-ATTACK, a novel framework that uses reinforcement learning to generate adversarial attacks against language models without needing access to internal model gradients. By learning from the target model's responses, this method achieves superior query efficiency and successfully bypasses safety filters by forcing unintended outputs.

arXiv cs.CL
Agents· Important· 🧪 Worth Testing

NVIDIA Enhances Agentic Security and Expands Korean Model Capabilities

NVIDIA has introduced the Open Agent Safety Platform, featuring the OpenShell runtime to provide sandboxed execution and policy-based governance for enterprise AI agents. Concurrently, LG AI Research has opened API access to K-EXAONE-2.5, a model supported by the new Nemotron-Personas-Korea dataset consisting of 7 million synthetic personas.

NVIDIA
Cyber Security· Important· 🧪 Worth Testing

Cybercriminals leverage ARTEX AI framework for financial sector attacks

The threat actor Storm-3358 is increasingly using the ARTEX agentic AI framework to accelerate attacks against financial organizations. The framework automates vulnerability scanning, credential harvesting, and lateral movement at machine speed, utilizing common tools like Nmap and curl.

Microsoft Security
Automation· Important· 🧪 Worth Testing

Amazon SageMaker now supports Multi-Instance GPU for managed endpoints

Amazon SageMaker Inference has introduced support for NVIDIA Multi-Instance GPU (MIG), allowing you to partition a single physical GPU into multiple independent units. This hardware-level division provides dedicated compute and memory to each partition, ensuring predictable latency and strong isolation between models hosted on the same accelerator.

AWS Machine Learning
Agents· Important· 🧪 Worth Testing

Scaling production AI agents with Amazon Bedrock AgentCore

Taxbit has deployed a secure, role-aware AI assistant using Amazon Bedrock AgentCore to automate product documentation support. The agent uses RAG to ground answers in verified content while strictly enforcing PII masking, regional data residency, and user-role permissions.

AWS Machine Learning
Agents· Important· 🧪 Worth Testing

Scaling autonomous security operations with Amazon Bedrock AgentCore

Greenlight Financial has successfully automated over 90% of its security and risk case resolution—totaling 4,000 cases monthly—by deploying a tiered multi-agent architecture on Amazon Bedrock AgentCore. The system uses a Tier 1 supervisor for high-volume triage and a Tier 2 supervisor for deep-dive investigation, significantly reducing human analyst workload.

AWS Machine Learning