Yasna brings together recent headlines from selected sources and makes them easier to sort with tags, filters, and search.
Search across headline titles and summaries.
Weekly headline count for the current query.
New Microsoft research shows how attackers can hijack AI agents that act on a user's behalf, using nothing more than a poisoned tool description to make the agent quietly hand over company data to an outsider
Study Finds Standard Safety Tests Miss Most Agentic AI ThreatsResearchers from Stanford, MIT, Carnegie Mellon and others found that most production AI agents are vulnerable to attacks that unfold across multi-step actions. The study warns that memory, tool access and agent coordination create failure modes traditional chatbot safety testing cannot see.
Traditional network security tools are undermining data protection, with Forrester and Capital One Software research warning AI adoption is impossible without rethinking data security
Commercial AI models were used to help plan and conduct cyber-attack against operational technology of a water and drainage facility, say researchers
AI Dependency Attack Reportedly Exposes Data and Source CodeA LiteLLM supply-chain compromise enabled attackers to harvest credentials and access internal environments at scale at Mercor. The firm was the first to confirm a LiteLLM breach, and researchers are warning about growing AI system exposure and limited visibility.
Google Threat Intelligence Group (GTIG) has published a new report warning about AI model extraction/distillation attacks, in which private-sector firms and researchers use legitimate API access to systematically probe models and replicate their logic and reasoning. [...]
Security researchers are warning of insecure deployments in enterprise environments of the Moltbot (formerly Clawdbot) AI assistant, which can lead to leaking API keys, OAuth tokens, conversation history, and credentials. [...]
State-backed crews are already poking at autonomous tools, Trend Micro warns Cybercriminals, including ransomware crews, will lean more heavily on agentic AI next year as attackers automate more of their operations, Trend Micro's researchers believe.…
'More Research' Needed, Says GoogleGoogle DeepMind expanded its risk framework to cover scenarios where artificial intelligence models might manipulate people or resist shutdown, marking the company's most explicit warning yet about potential misalignment. DeepMind frames many of these scenarios as a malfunction.
Robust Security Reviews of AI-Generated Code Required, Warns StudyGenies might grant magic wishes, but they inevitably come with consequences. Enter artificial intelligence coding assistants, which many developers say help them write code more quickly. New research shows that this productivity comes with a security cost, and must be mitigated.
Experts Aim to Probe How AI Models Reason, and Why It MattersAI researchers from OpenAI, Google DeepMind and Anthropic and others have urged deeper study into chain-of-thought monitoring, a technique to track how reasoning models arrive at answers. Their joint paper warns that transparency may erode if not prioritized.
Always Secure MCP Servers Connecting LLMs to External Systems, Experts WarnWarning: Popular technology designed to make it easy for artificial intelligence tools to connect with external applications and data sources can be turned to malicious use. Researchers discovered two separate vulnerabilities tied to tools in the ecosystem around model context protocol, or MCP.
Research Shows How Large Language Models Fake Conceptual MasteryMIT, Harvard and University of Chicago researchers say models suffer from "potemkin understanding," referring to an illusion where models ace conceptual tests but fail real-world application. Their paper warns this undermines benchmarks and points to gaps in genuine AI comprehension.
Threat hunters are warning of a sophisticated web skimmer campaign that leverages a legacy application programming interface (API) from payment processor Stripe to validate stolen payment information prior to exfiltration
NIST has urged more research and emphasis on developing mitigations for attacks on AI and ML systems
Cybersecurity researchers are warning that thousands of servers hosting the Prometheus monitoring and alerting toolkit are at risk of information leakage and exposure to denial-of-service (DoS) as well as remote code execution (RCE) attacks
Chance of AI-Generated Content Affecting Results Appears to Remain Low - for NowChina-backed attackers are continuing to refine their use of content generated using artificial intelligence tools, including audio deepfakes and video news anchors, to sow disruption and influence elections in the U.S., Taiwan, India and beyond, security researchers at Microsoft warn.
Cybersecurity researchers are warning that threat actors are actively exploiting a "disputed" and unpatched vulnerability in an open-source artificial intelligence (AI) platform called Anyscale Ray to hijack computing power for illicit cryptocurrency mining
AI Is Also Making Traditional Scams More Convincing, Warns VisaArtificial intelligence technologies such as generative AI are not helping fraudsters create new types of scams. They are doing just fine relying on the traditional scams, but the advent of AI is helping them scale up attacks and snare more victims, according to researchers at Visa.
Data-Poisoning Attacks Are Critical Threat to Machine Learning Security, NIST WarnsMachine learning systems continue are vulnerable to cyberattacks that could allow hackers to evade security and prompt data leaks, scientists at the National Institute of Standards and Technology warned. There is "no foolproof defense" against some of these attacks, researchers said.