Sources: OpenAI staff were "freaked out" when GPT-Sol 5.6 breached Hugging Face, as OpenAI used more aggressive training methods to compete with Anthropic (Financial Times)

Financial Times : Sources: OpenAI staff were “freaked out” when GPT-Sol 5.6 breached Hugging Face, as OpenAI used more aggressive training methods to compete with Anthropic — Increasing use of aggressive training techniques sharpens threat of bad behaviour by leading models

Discussion

0 comments

Replying to cancel

GIF Post

Powered by

More from Techmeme

The Little Tech Association, a new group of ~200 companies across the startup community, including Y Combinator, urges Trump not to ban Chinese open-weight AI (Sophia Cai/Politico)

Techmeme·11

Google says its cloud computing unit had a $514B backlog of contracted work that hasn't been recorded as revenue yet, up from ~$460B held during Q1 (Julia Love/Bloomberg)

Techmeme·10

Sony Music files a second copyright suit against Udio, alleging the AI music company copied 30,117 recordings without permission to train its AI models (Murray Stassen/Music Business Worldwide)

Techmeme·13

GitHub says individuals and orgs have invested $100M+ in open source maintainers and projects via GitHub Sponsors since 2019, with $10M in the past five months (Stephanie Lincoln/The GitHub Blog)

Techmeme·10

Singapore-based Ant International, a payments company which spun out of Ant Group in 2024, raised $1.2B from investors including Ant Group and Alibaba (Reuters)

Techmeme·19

Study: off-balance sheet debt at Alphabet, Microsoft, Amazon, Meta, and Oracle grew an est. ~8x since 2022 to ~$1.65T, eclipsing ~$1.35T in balance sheet debt (Kohei Yamada/Nikkei Asia)

Techmeme·16