Google Genie 4: Steerable AI Worlds, Audio and the Limits of Its Preview
Google Genie 4 creates interactive worlds with audio and real-time steering. Who can test it now, what can it do, and where does it still fail?
Google Genie 4 creates interactive worlds with audio and real-time steering. Who can test it now, what can it do, and where does it still fail?
Anthropic’s IPO filing gives its benefit trust a board majority and founders voting power on other matters. A later staggered board plan raises accountability questions.
Cohere says a Qwen AI agent tried to upload a placeholder string during a security test. The sandbox blocked it; the cause and intent remain unknown.
Block demonstrated a research voice model that can listen while speaking and generate expressive speech. Dessa is not publicly available, and its performance claims await independent testing.
OpenAI says its internal frontier-model training and most evaluations will no longer use the live internet. Consumer ChatGPT browsing is not the subject of this policy.
MIT-led researchers let 13 AI agents explore a virtual island for 30 hours. Their behavior resembled play, but the study does not prove sentience.
AI-assisted researchers found critical PHP flaws, and patches are out. Here is what the 70% website statistic does—and does not—mean.
ChatGPT Locked Chats can hide a conversation from casual viewing, but the PIN does not change encryption, retention, memory already stored, or model-improvement settings.
AI helped historians analyze 100,000 pages of port books. The records reveal surprising European trade links, lost ports and unusual cargo.
Reuters reports more than 1,200 AI-assisted intrusion attempts and over 100 reportedly compromised sites. Here is what the researcher estimated—and what remains unverified.