News
MCP Python SDK advisories expose task ownership and browser-origin gaps
Two narrowly scoped advisories show why agent control planes still need ordinary session authorization and transport security.
AI security research / news
News desk
Reports, updates and corrections.
95 reports and updates · Page 4 of 4
5 matching records on this page
Published news
News
GPT-Red turns model robustness testing into an automated adversarial loop
OpenAI reports an automated red-team loop that improved attack success against indirect prompt-injection defenses in a controlled arena. The result is benchmark evidence, not a production compromise.
News
GhostCommit hides coding-agent instructions in PNGs that text-only review can miss
GhostCommit put instructions in a repository image that a coding agent later followed, copying synthetic secrets into generated code. The controlled research shows an input-review blind spot and an output channel, not confirmed theft from production repositories.
News
Check Point turns an AI-generated malware idea into a browser-only ransomware proof of concept
A controlled Android demonstration used legitimate File System Access permissions to encrypt selected images without a native payload.
News
CyberChainBench grades exploit agents against historical chain state and economic impact
CyberChainBench evaluates agents against 541 historical EVM incidents across nine chains. Its best setup achieved 43.7% exploitation on the evaluated set, reproducing historical profit in forked state rather than stealing live funds.
No matches on this page.