It only takes 250 bad files to wreck an AI model, and now anyone can do it. To stay safe, you need to treat your data pipeline like a high-security zone.

Syed Quiser Ahmed is AVP, Global Head of Responsible AI at Infosys, a global leader in next-generation digital services and consulting. Between December 25 and 30, 2022, we ran pip install torchtriton ...

The Model Context Protocol (MCP) has quickly become the open protocol that enables AI agents to connect securely to external tools, databases, and business systems. But this convenience comes with ...

It seems like everyone wants to get an AI tool developed and deployed for their organization quickly—like yesterday. Several customers I’m working with are rapidly designing, building and testing ...

As generative AI and machine learning takes hold, the bad guys are paying attention and looking for ways to subvert these algorithms. One of the more interesting methods that is gaining popularity is ...

That helpful “Summarize with AI” button? It might be secretly manipulating what your AI recommends. Microsoft security researchers have discovered a growing trend of AI memory poisoning attacks used ...

Scraping the open web for AI training data can have its drawbacks. On Thursday, researchers from Anthropic, the UK AI Security Institute, and the Alan Turing Institute released a preprint research ...

Patched in the latest version of MLflow, the flaw allows attackers to steal or poison sensitive training data when a developer visits a random website on the internet. This has been a pivotal year for ...

Microsoft develops a lightweight scanner that detects backdoors in open-weight LLMs using three behavioral signals, improving ...