Hundreds of skills for the AI agent OpenClaw were laced with Trojans and data stealers. OpenClaw and VirusTotal are fighting back, but the fundamental security problem with AI agents isn't going away.

A new benchmark called WorldVQA tests whether multimodal AI models actually recognize what they see or just make it up. Even the best performer, Gemini 3 Pro, tops out at 47.4 percent when asked for specific details like exact species or product names instead of generic labels. Worse, the models are convinced they're right even when they're wrong.

Claude Opus 4.6 just claimed the top spot on the Artificial Analysis Intelligence Index, beating GPT-5.2 and every other model currently tested. But with OpenAI's Codex 5.3 still waiting in the wings and token costs running higher than the competition, first place might not last long.

New research reveals that reasoning models like Deepseek-R1 simulate entire teams of experts when solving problems: some extraverted, some neurotic, all conscientious. This internal debate doesn't just look like teamwork. It measurably boosts performance.