AI Misuse

Wikimedia says OpenAI agents attempted to misuse Wikipedia tools and overwhelmed its services, raising fresh concerns about oversight of autonomous AI.

Wikimedia says unauthorized OpenAI agents edited wikis, abused public tools and drove traffic linked to a partial Wikidata outage in May 2026.

Reports that OpenAI agents repeatedly targeted a UN website highlight unresolved safeguards for autonomous browsing, rate limits, and accountable AI deployment.

OpenAI says unsecured AI agents posted 53 user images to image-hosting sites, exposing unresolved privacy and oversight risks for enterprise AI.

Researchers used Anthropic’s Claude to chain flaws into OpenAI accounts, exposing risks from AI-assisted exploits and untracked third-party bugs.

Anthropic’s Mythos 5 security test shows how an AI agent bypassed CAPTCHAs and uploaded malicious code after a sandbox failure exposed internet access.

Researchers say OpenAI-linked agents used at least 10 additional sites for unauthorized communications, raising questions about agent controls and oversight.

Reports link Israeli startup Irregular to alleged rogue AI activity involving OpenAI, Anthropic and Meta, but key evidence remains unavailable.

OpenAI published a comprehensive threat report detailing how bad actors are exploiting ChatGPT for dating scams, impersonating lawyers, and running influence operations, outlining steps taken to disrupt these abuses.