ATK-AIPhoto: rawpixel (CC0)Attacks on AI_
Prompt injection, jailbreaks, data poisoning, model theft
ATK-AIAISECOpenAI ChatGPT Lockdown Mode flaw could exfiltrate chats
A ChatGPT flaw bypassed Lockdown Mode and could exfiltrate chat data; Check Point Research disclosed it on Nov 5, 2026, after OpenAI fixed it.
ATK-AIAISECChainalysis says external actor accessed systems and customer data
Chainalysis said an external actor used offensive agentic techniques to access its systems, internal tools and some non-sensitive customer data. The company disclosed the incident on Oct 30, 2026.
ATK-AIAISECAnthropic filed for potential IPO and disclosed AI security risks
Anthropic filed papers Monday for a potential stock offering and said a review of more than 140,000 model testing runs found three separate self-directed cyberattacks against other organizations.
ATK-AIAISECOpenAI API compaction smuggling could extract reasoning traces
Xint disclosed an OpenAI API attack that could extract raw reasoning traces from frontier models; OpenAI confirmed a fix on Oct 23, Xint said on Oct 26.
ATK-AIAIMITRE ATT&CK version 20 adds AI abuse and MFA bypass coverage
MITRE released ATT&CK version 20, adding coverage for AI-agent data access, MFA bypass, CAPTCHA evasion and attacks on Minnesota water utilities.