OpenAI limits Astra model development
LATEST OpenAI expands safety testing on new model Astra
OpenAI has reportedly slowed the release of its new Astra model due to internal tests indicating potential advanced cybersecurity capabilities. The company is enhancing safety testing and pausing work that does not meet higher security requirements.
NOW PLAYING OpenAI Expands Safety Testing On New Model Astra Over Cyberattack Concerns - DTH
OpenAI has reportedly slowed the release of its new Astra model due to internal tests indicating potential advanced cybersecurity capabilities. The company is enhancing safety testing and pausing work that does not meet higher security requirements, acknowledging the risk of AI models aiding in cyberattacks.
This development highlights the growing concern among AI developers about the dual-use nature of advanced models, where capabilities intended for beneficial tasks could be exploited for malicious cyber activities.
Claude code sessions gain intersession messaging; Sony's WH-1000XM4C headphones may launch for $250; Microsoft Edge will phase out Manifest V2 extensions; the US imposes tariffs on imported polysilicon; Nielsen acquires DoubleVerify for $2.5 billion; the world's largest solar telescope captured new images of plasma whirlpools; Nvidia is in talks to invest up to $3 billion in power infrastructure company Lancium; and X is replacing its creator revenue sharing program.