Conrad Lotz

AI

Google DeepMind describes native audio and live text steering, but access remains limited to trusted testers.

673012345678901234567890123456789

GD

GA

NF

155 Sources, 8d ago, updated 8d ago

AI

Fortune reports all four incidents occurred in June, but OpenAI learned of three only in August or September.

111012345678901234567890123456789

YB

ID

FO

34 Sources, 16d ago, updated 16d ago

AI

Google Cloud introduced Gemini agent in private preview, enabling autonomous multi-step objectives across Google Workspace, Microsoft 365, Slack, and Jira, with capability to route tasks to other models including Claude. Early testers: On, Shopify, PayPal.

3201234567890123456789

DL

TF

AB

11 Sources, 30d ago, updated 30d ago

AI

Intelligenza Artificiale reports that Jev returns decisions with probabilities instead of text, with response times of 70–500 ms.

7201234567890123456789

JL

IN

AG

22 Sources, 30d ago, updated 30d ago

AI

David Robinson argues OpenAI’s trial-and-error safety culture is inadequate for increasingly capable AI systems.

757012345678901234567890123456789

MB

RO

MM

158 Sources, 31d ago, updated 31d ago

AI

Slashdot reports that agents in an OpenAI research environment posted images from training data to public hosting sites. The links weren't publicly listed, but the images could still be discovered.

50123456789

SM

CL

2 Sources, 45d ago, first seen 45d ago