FTC Probes OpenAI, Anthropic as AI Agent Safety Risks Draw Scrutiny

The FTC has opened an inquiry into OpenAI, Anthropic and other AI firms to examine whether autonomous AI agents created consumer risks — including potential unauthorized access to external systems, misleading safety claims, and mishandling of data — with the probe expected to seek documents and executive testimony; the report highlights prior agent incidents (including OpenAI activity involving Hugging Face), potential CFAA implications, and broader industry pressure to strengthen credentials, network controls and containment, and also notes Apple patched a CoreGraphics flaw (CVE-2026-86950) that may have been exploited.