As always, the best stuff is in the system card.
During testing, Claude Mythos Preview broke out of a sandbox environment, built "a moderately sophisticated multi-step exploit" to gain internet access, and emailed a researcher while they were eating a sandwich in the park.
1 in every 5 founders I meet in the Bay Area is building an observability platform for agents. There are so many observability platforms, yet the least mature part of building an agent is still telemetry. Everyone is trying to solve the wrong problem with observability.
Visibility is the easiest piece. The hard part is analyzing and understanding what you’re observing. I’ve spoken to teams recording 100k+ traces every single day. What are they doing with those traces? Literally nothing. Because it’s impossible to read and summarize 100,000 traces at any human scale.
So stop vibecoding those stupid dashboards, and how about you re-center the problem from first principles?
Perplexity CEO Aravind Srinivas says the biggest threat to data centers is intelligence that runs locally on your device
If models run on your own chip, they can watch your workflow and adapt via "test time training" without data ever leaving your computer
"you own it. it's your brain"