Anthropic published the prompt injection failure rates that enterprise security teams have been asking every vendor for

Anthropic's Opus 4.6 system card breaks out prompt injection attack success rates by surface, attempt count, and safeguard ...
The Hacker News

RoguePilot Flaw in GitHub Codespaces Enabled Copilot to Leak GITHUB_TOKEN

RoguePilot flaw let GitHub Copilot leak GITHUB_TOKEN, while new studies expose LLM side channels, ShadowLogic backdoors, and promptware risks.
AOL

OpenAI’s new safety tools are designed to make AI models harder to jailbreak. Instead, they may give users a false sense of security

OpenAI last week unveiled two new free-to-download tools that are supposed to make it easier for businesses to construct guardrails around the prompts users feed AI models and the outputs those ...