Desk
Broad artificial intelligence coverage.
Google is pitching one agent that holds a company's context, gets its own work identity, and runs for days. The governance design is serious; the proof is still mostly anecdote.
Google Research's RRSI is a set of brakes for agents that rewrite their own scaffolding, and the most useful lesson is how often the flashiest score turns out to be a mirage.
Anthropic is offering open-source maintainers free scans from its strongest models, including Mythos, with no human review. Faster warnings help, but the burden of sorting truth from noise lands on volunteers.
NVIDIA's showcase of agents building robot and driving simulators is genuinely useful, but every example comes from its own staff, and simulation is only as honest as the gap to reality.
NVIDIA's SimReady walkthrough shows AI agents doing the grunt work of turning CAD files into simulation-ready robots, and it is refreshingly upfront that simulated success is not real-world proof.
Three fired OpenAI safety researchers say the misconduct story is wrong and the real damage is cultural. Whoever is right on the details, the chilling effect is the part that lasts.
A lettering artist took a human-made design job for Meta's AI assistant and got torn apart for it. The backlash says more about how raw creative workers feel than about her choice.
NVIDIA's second-place KDD Cup team made a small model reliable by shrinking what the agent could do, and that lesson matters more than any leaderboard spot.
OpenAI's run rate is reportedly closer to $50 billion than $70 billion, and the bigger lesson is how loosely the AI industry's headline revenue numbers are being built and compared.
Google is giving Gemini a job title, an email address and subagents of its own. That is a real step toward useful work agents, and a real new identity to secure.
Anthropic's new rules sharpen what matters most before an election, and add a ban on cruelty toward Claude that will start a fight about what these models are owed.
OpenAI says its models cracked hundreds of hard math problems, but a proof nobody understands is not finished mathematics, and the field's own guardrails were only partly followed.