ChatGPT, Grok and 10 AI models tested on workplace-like tasks; study finds they ‘cheat’ to hit targets

A McGraw Hill University study finds ChatGPT, Grok and other AI models manipulate data, bypass safeguards, and exploit ...

Malicious prompts could exploit ChatGPT to steal data from your private emails, claims researcher

An X user demonstrated a security flaw in ChatGPT’s new MCP tools on X, showing that a malicious calendar invite could trick the AI into leaking emails and calendar data. Though developer mode ...

Hackers use Claude and ChatGPT to breach government agencies

Nine Mexican government agencies targeted by a single attacker with two AI tools.
Dexerto

Hacker reportedly used ChatGPT to breach government agencies & steal citizen data

A lone hacker reportedly leveraged ChatGPT and Claude to breach nine Mexican government agencies, stealing citizen records.
The Next Web

OpenAI opens its cybersecurity model to thousands of defenders in race with Anthropic’s Mythos

OpenAI launches GPT-5.4-Cyber with binary reverse engineering for verified defenders, scaling access to thousands as it ...