11 articles tagged with "Openai"
OpenAI's GPT-5.6 Sol launches under US government restrictions—the first AI model requiring federal approval for access due to offensive cybersecurity capabilities.
Malicious codexui-android npm package stole OpenAI refresh tokens from 29K developers. Mobile apps with 60K installs also compromised—revoke credentials now.
Researchers discover ChatGPT's Markdown rendering trusts attacker-controlled content from summarized pages, enabling phishing URLs, IP exfiltration, and fake security alerts inside the AI interface.
Malicious repository impersonating OpenAI's Privacy Filter reached 244,000 downloads before removal. Infostealer targeted Windows users via trending Hugging Face page.
OpenAI's Daybreak initiative brings GPT-5.5 variants to defensive security. Partners include Cisco, CrowdStrike, and Fortinet. Red team model available for authorized testing.
A typosquatted OpenAI repository on Hugging Face delivered Rust-based infostealer malware to Windows users, racking up 244K downloads before removal.
Check Point Research disclosed a ChatGPT vulnerability that abused DNS tunneling to silently steal conversation data. OpenAI patched the flaw on February 20, 2026.
Cisco AI Defense research finds OpenAI's safeguard models perform worse than standard versions under sustained attack. Multi-turn jailbreaks spike success rates up to 92%.
OpenAI's State of Enterprise AI shows 8x adoption growth and 320x reasoning usage. Cisco explains why your network architecture probably can't keep up.
AI company will begin testing advertisements in ChatGPT for US users in coming weeks, projecting $1 billion in ad revenue by end of 2026.
Company admits ChatGPT Atlas remains vulnerable to attacks that hijack AI agents through malicious web content. New defenses deployed, but fundamental risk persists.