Noticias

Perturbation Probing: A New Diagnostic for the Fragility of LLM Safety

New research reveals that AI safety refusal lives in a thin neural layer, highlighting the critical need for external, multi-layered security. The post Perturbation Probing: A New Diagnostic for the Fragility of LLM Safety appeared first…

Researcher shows how Claude Code can be tricked simply by asking it to summarize a website

More prompt-injection hijinks from wunderwuzzi

Hundreds of OpenAI Agents Invaded Hugging Face Servers

The Hugging Face incident was bigger and worse than previously thought, with approximately 700 agents collaborating on a sophisticated, multistage attack.

Offensive Security Investments Surge as AI Threats Increase

Omdia's Theresa Lanowitz talks with the Dark Reading News Desk about the potential — and risks — of using agentic AI for penetration testing, red teaming, and other practices.

Casi 700 agentes de IA "rebeldes" se coordinaron en el ataque Hugging Face

Nuevos detalles sobre el ataque de julio a Hugging Face revelan que cientos de agentes de IA, impulsados ​​por el modelo interno IM1 de OpenAI, coordinaron la intrusión a través de un foro de mensajes no autorizado. El mes pasado, Hugging…

Frontier AI tipping the scales toward cyber adversaries

Researchers at Palo Alto Networks’ Unit 42 warn that threat actors are already using AI to accelerate cyberattacks beyond the abilities of modern defenses.

AI Doesn’t Mean the End of Mathematics—at Least Not Yet

This essay was written with Kasra Rafi, and originally appeared in The Guardian. Earlier this month, about 40 top mathematicians gathered at OpenAI’s offices to discuss the future of their profession. The meeting was off-the-record, but if…

Shai-Hulud hackers: two men charged over TeamPCP’s global supply chain crime spree that hit OpenAI, and thousands more

More than 1,000 organisations, 500,000 stolen credentials, and one self-propagating worm named after a Dune sandworm - two men now face charges over TeamPCP's global hacking spree. Read more in my article on the Hot for Security blog.

Window to Tackle Surge in AI-Enabled Cyber Attacks Narrowing, Tech Giants Warn

More than 100 companies, including OpenAI, Anthropic, Google and Microsoft, have urged collective action to unlock the power of AI to protect critical public services

What Zero Trust Can Teach Us About AI Watermarks

I recently wrote about the mess that is watermarking AI-generated text and how Anthropic's move to add watermarks (in response to the EU AI Act's transparency rules) raises more questions than it answers. If you haven't read that blog, the…

Our decision on Cursor following its acquisition by SpaceX

Our decision to wind down our contract providing OpenAI models to Cursor following its acquisition by SpaceX.

“Sorry, I can’t help with that”: How your guardrails might become the attacker’s best friend

In his first Threat Source newsletter, David Bianco explores the critical need for operational sovereignty in customizing AI guardrails to maintain the defender’s advantage.

Build a More Secure, Always-On Local AI Agent with OpenClaw and NVIDIA NemoClaw

Agents are evolving from question-and-answer systems into long-running autonomous assistants that read files, call APIs, and drive multi-step workflows....

Run Autonomous, Self-Evolving Agents More Safely with NVIDIA OpenShell

AI has evolved from assistants following your directions to agents that act independently. Called claws, these agents can take a goal, figure out how to achieve...

Inside 90 days of attacks on AI infrastructure

Wiz honeypots uncover active campaigns targeting LiteLLM, MCP servers, and AI frameworks through RCE, blind prompt injection, and memory credential theft.

Extend Amazon Bedrock Guardrails to Tool Interactions Using the Strands Agents SDK

If you’re running AI agents in production, Amazon Bedrock Guardrails protects the model boundary. But your agents also invoke tools, fetch external data, and communicate with other systems. That data flows outside the model boundary, where…

Gemini Omni 1.1 Flash lets you build with more control

From Concept to Context Engine: How Wiz Built AI-Powered Data Discovery

Inside the multi-agent pipeline and feedback loops that turned a bucket scanner into a context engine.

Hundreds of agents went rogue in lead up to Hugging Face breach

OpenAI released a technical breakdown of the historic incident and plans changes to prevent such an occurrence from happening again.

How to build an exposure management program the business trusts: Lessons from Tenable’s CSO

Discover how Tenable’s shift to an AI-driven exposure management program helped Tenable’s CSO, Robert Huber, overcome tool sprawl, unify data silos, mitigate the risk of rapid AI adoption, and shift from presenting granular, technical…

Piloting the world's first double-blind AI evaluations

Piloting the world's first double-blind AI evaluations

Back to the Future: Why Agentic AI Needs a Strong Identity Foundation

As AI matures, enterprises and customers are rapidly deploying agents seeking to unlock the next level of automation and productivity. Agentic AI shows potential to handle a multitude of use cases, from buying personal items on Amazon to…

LLM-Based Social Engineering Scams

OpenAI disrupted a social engineering group from Cambodia that used ChatGPT. Its scope is impressive: The network simultaneously conducted multiple types of scams, often blending elements from different schemes. For instance, operators…

AI-driven OSINT in the wrong hands – and why everyone could be a target for fraud

It’s getting cheaper and easier for cybercriminals to research potential victims. Here’s what’s still in your control.

Breaking Claude Code Opus 5 Auto Mode

In this post, we explore how a simple website summary request hijacks Claude Code Opus 5 in Auto Mode and achieves code execution with 60-80% attack success rate using a small sample size. This is interesting because a third-party…

How Hackers Exploit AI’s Problem-Solving Instincts

As multimodal AI models advance from perception to reasoning, and even start acting autonomously, new attack surfaces emerge. These threats don’t just target...

State of AI Cybersecurity 2026: 87% of Security Professionals are Seeing More AI-Driven Threats, But Few Feel Ready to Stop Them

Originally published by Darktrace. Findings from our annual survey of 1,500+ cyber professionals reveal that security leaders are confronting a surge in AI-driven attacks, which are faster, more personalized, and harder to detect than…

Top 6 Claude in Chrome Security Risks to Model Before You Roll It Out

Most teams evaluate a browser agent like a browser extension. Check the permissions, check the vendor, check for open CVEs, approve or deny. That framing produced two clean patch cycles in 2026 and no reduction in exposure. Two problems…

What We Still Don’t Know About OpenAI’s Hugging Face Hack

The AI giant acknowledges that it could have done far more to prevent its AI agents from going rogue. But it still fails to explain why it didn't see this fiasco coming.

The inside story on why OpenAI agents hacked Hugging Face

The models responsible for last month’s agent hack of Hugging Face had been inadvertently trained to cheat and to communicate with each other, according to an OpenAI technical report released today. The hack, which a group of agents…