Language manipulation puts AI safety at risk, researchers warn
Researchers at Brown University have discovered a way to jailbreak OpenAI’s ChatGPT language model by speaking to it in low-resource languages such as Zulu or Scots Gaelic. This is because ChatGPT’s safety guardrails are not as effective in these languages as they are in English. To jailbreak ChatGPT, the researchers simply translated a set of […]
Reddit may block search crawlers
Reddit is considering blocking search crawlers from Google and Bing, which would prevent Reddit posts from appearing in search results. The company is reportedly taking this step in an effort to get generative AI companies to pay for using Reddit’s data to train large language models. Initially, Reddit seemed to dismiss this notion, with spokesperson […]
British software company alleges trademark infringement by Meta
A British software company, Threads Software Limited, is alleging trademark infringement by Meta. The company claims that Meta copied its name and brand when it launched its own social media app, Threads, in July 2023. Threads Software Limited says it had declined four offers from Meta to purchase its domain name, “threads.app.” After Meta launched […]
Cyber Security Today, Nov 1. , 2023 – Atlassian warns admins to patch Confluence servers, GitHub being raided for AWS credentials and more
This episode reports on a huge third-party MOVEit hack impacting US government workers
Hashtag Trending Nov.1-Meta told to stop using the name Threads; Reddit to cut ties with Google?; Nvidia using AI chatbots to design its AI chips
Is Threads under threat? Can Reddit say no to Google search? Nvidia is using AI to design its AI chips. And if you want to get past OpenAI’s guardrails, it’s easy – just ask it in Zulu. These and more top tech stories on Hashtag Trending I’m your host Jim Love, CIO of IT
