Docker outlines the main security risks of autonomous AI agents, including prompt injection, excessive autonomy and exposed credentials. The guide also covers ways to contain those risks.
The Harbor evaluation tool can now run each eval task inside Docker Sandboxes, microVM environments built to run agents such as Claude Code, Codex and Cursor safely. Custom network policies help keep evaluations fair and repeatable, either locally or in the cloud.
At AI Engineer New York, frontier labs and financial firms discussed how much autonomy to give AI agents in finance. The focus was on keeping consequential decisions under control while still letting agents do useful work.
Eight US states have taken OpenAI to court over concerns about ChatGPT's safety. The lawsuits add to growing scrutiny of the company's safety and security practices, particularly how it protects children.
Gudea, which calls itself a "storm tracker for the internet", follows how stories and rumors spread online. The company aims to spot posts involving AI-generated misinformation, hate speech or damaging rumors before they grow into public controversies.
Bank of America brought in 1,000 contractors to fix software vulnerabilities uncovered by Anthropic's Mythos model. The move offers a rare glimpse of what defending against AI-discovered flaws costs a large company.