Tag: AI Agents
Google and Nvidia both decided agents need their own plumbing
Within hours of each other, Google shipped a microVM per agent and Nvidia shipped a telemetry layer for…
Uber hands an office agent to thirty thousand employees
A thousand-person pilot that skewed heavily non-engineer convinced Uber to make ChatGPT Work standard kit for its whole…
Sakana packages its consulting work into an agent platform
Aquarium puts an ontology, approval rules, and a managed runtime under one roof so enterprise agents can share…
Outside testers cracked OpenAI’s agent watchdog three ways
A three-week embedded review found that OpenAI's internal misalignment monitor could be dodged, blinded, or made to crash.
Trail of Bits finds its own patches broke working code
After letting agents write security fixes, the firm audited what it had merged and confirmed 21 regressions across…
1Password lets an agent log in without ever holding the key
The password manager now fills in credentials an AI assistant requests, so the model completes the task but…
CyberCartel ran a four-model debate before robbing a bank
Straiker says it read the criminal group's own orchestration files and found Claude agents arguing over targets while…
OpenAI buys Orchestra Bio to manage the lab work AI cannot do
OpenAI has absorbed a drug-development coordination startup and folded it into Rosalind, its life-sciences model family.
IBM gets top customer service scores from notes, not new weights
ALTK-Evolve distills an agent's successes and failures into written rules, letting an older model beat newer entries on…
Meta suspends Muse sharing after researchers enumerate 6,800 links
A guessable URL scheme exposed conversations, images and business details that users believed were private.
Bob McGrew’s Arda raises $230M to put agents on factory floors
The former OpenAI research chief's startup is valued at $1.5B for agent software that rebuilds assembly lines on…
Microsoft turns its security agents inward on the agents it sold
Guardian Agents watch enterprise deployments for insider-style behaviour, built on the premise that frontier models cannot be trusted.