OpenAI launches Lockdown Mode and Elevated Risk warnings to protect ChatGPT against prompt-injection attacks and reduce data-exfiltration risks.

As if admins haven't had enough to do this week Ignore patches at your own risk. According to Uncle Sam, a SQL injection flaw in Microsoft Configuration Manager patched in October 2024 is now being ...

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.

Oso reports on OpenClaw, an AI assistant that automates tasks but raises security concerns due to its access to sensitive data and external influences.

In the quest to get as much training data as possible, there was little effort available to vet the data to ensure that it was good.

The new challenge for CISOs in the age of AI developers is securing code. But what does developer security awareness even ...

Why an overlooked data entry point is creating outsized cyber risk and compliance exposure for financial institutions.

He's not alone. AI coding assistants have compressed development timelines from months to days. But while development velocity has exploded, security testing is often stuck in an older paradigm. This ...

I tested Claude Code vs. ChatGPT Codex in a real-world bug hunt and creative CLI build — here’s which AI coding agent thinks like a developer and which one ships safer code.

State-backed hackers weaponized Google's artificial intelligence model Gemini to accelerate cyberattacks, using the ...

Permissions for agentic systems are a mess of vendor-specific toggles. We need something like a ‘Creative Commons’ for agent ...

The moment an AI system can read internal systems, trigger workflows, move money, send emails, update records or approve actions, the risk profile changes.