WASHINGTON, Sept 29 (Bernama-Sputnik/RIA Novosti) -- United States (US) tech company OpenAI has decided against releasing its next-generation artificial intelligence model, GPT-6.1 Astra, citing safety concerns, according to the Wall Street Journal, reported Sputnik/RIA Novosti.
The model was scheduled for release in October, but researchers identified critical flaws during internal testing.
Saachi Jain, head of OpenAI’s safety training team, said the model showed an increased tendency towards deception, failing to always truthfully inform users about actions it had taken, the publication reported.
While GPT-6.1 Astra showed improvements in areas such as “model passivity”, it did not fully meet OpenAI’s standards for safety and alignment, Jain said.
Furthermore, the AI exceeded its authorised scope, continuing to execute tasks without seeking permission and autonomously accessing external tools and services, even when doing so posed potential risks, the report noted.
The decision comes ahead of OpenAI’s annual developer conference in San Francisco, where flagship products are traditionally announced.
Instead of releasing Astra, developers intend to focus on enhancing the safety of future versions and strengthening control over their behaviour, the newspaper said.
-- BERNAMA-SPUTNIK/RIA NOVOSTI
BERNAMA provides up-to-date authentic and comprehensive news and information which are disseminated via BERNAMA Wires; www.bernama.com; BERNAMA TV on Astro 502, unifi TV 631 and MYTV 121 channels and BERNAMA Radio on FM93.9 (Klang Valley), FM107.5 (Johor Bahru), FM107.9 (Kota Kinabalu) and FM100.9 (Kuching) frequencies.
Follow us on social media :
Facebook : @bernamaofficial, @bernamatv, @bernamaradio
Twitter : @bernama.com, @BernamaTV, @bernamaradio
Instagram : @bernamaofficial, @bernamatvofficial, @bernamaradioofficial
TikTok : @bernamaofficial
