News - subcat

LLM-driven robots easy to jailbreak
Published in News
Monday, 25 November 2024 10:13

LLM-driven robots easy to jailbreak


Bake a file in a cake

A study has revealed an automated method to breach large language model (LLM)-driven robots with "100 per cent success" which can jail break a robot to turn it into a killing machine. 

MediaTek releases Dimensity 9400 chipset
Published in News
Wednesday, 09 October 2024 09:21

MediaTek releases Dimensity 9400 chipset


Based on TSMC’s 3nm process

MediaTek has lifted the kimono on its new Dimensity 9400 chipset, constructed on TSMC’s 3nm process, and featuring an All Big Core design

LLM’s have the same security problems as 1970s phones
Published in AI


John Draper could exploit it with his Captain Crunch whistle

Security expert Bruce Schneier has pointed out that large language models are open to the same kind of weaknesses that phones in the 1970s had, the kind that John Draper exploited.

ChatGPT is a terrible cancer doctor
Published in News
Tuesday, 05 September 2023 11:11

ChatGPT is a terrible cancer doctor


Does not do much else than stress you out

While ChatGPT might have the "brains" to pass a medical exam, a recent study published in the journal JAMA Oncology shows that the AI provides unhelpful information on cancer treatment, potentially harming patients who are already stressed out and looking for answers.

Brave search engine gets larger language model
Published in AI


Gets data from search results 

The people who bought you the Brave browser have improved its search engine to use a large language model to produce answers to your search queries.