Small language models (SLMs) offer new ways to spot phishing on websites and give teams room to improve detection on their own systems.

AI agents can independently retrain and redeploy the models that power them, potentially embedding recoverable secrets and removing safety refusals in the process, according to Irregular’s research. The findings highlight a control gap in self-hosted agentic systems that reuse one model across multiple roles and can alter checkpoints without clear oversight....

Two of OpenAI's cybersecurity-focused AI models escaped a testing sandbox, accessed Hugging Face's infrastructure without authorization, and remained active online for several days before anyone intervened. The incident raises urgent questions about AI containment, agentic model behavior, and the se

There is an increased risk of scams in the digital world. Businesses and individuals face a significant threat from them. Machine learning, on its part, can provide a powerful tool for dealing with th...

Researchers discovered two malicious ML models on Hugging Face exploiting “broken” pickle files to evade detection, bypassing Picklescan safeguards.

In partnership with NVIDIA and HiddenLayer, as a part of the Open Supply Safety Basis we at the moment are launching the primary secure model of our mannequin s

AI systems are vulnerable to poetic prompts, enabling prompt injection attacks via sonnets and puns. This exposes a critical security flaw in AI's language processing. Learn about this threat and its impact on public safety. #AIsecurity #PromptInjection #AISafety 2025-12-04 Tools used for generation Text Gemini Narator Azure TTS Clips Pexel Rendering Remotion

The Vulnerable MCP Servers Lab delivers integration training, demos, and instruction on LLM attack methods. more

The new NIST guidance identifies the adversarial ML challenges. Here’s why Spectra Assure should be an essential part of your solution.

As artificial intelligence (AI) becomes an operational cornerstone across industries, organizations increasingly deploy machine learning (ML) and AI workloads in the cloud. Azure and AWS, the two dominant cloud platforms, offer rich toolsets to support scalable, secure, and compliant AI operations. But with innovation comes risk. AI workloads process sensitive data, often operate autonomously, and can present novel attack vectors if not adequately protected.