Anthropic and other researchers detail “subliminal learning”, where LLMs learn traits from model-generated data that is semantically unrelated to those traits
We study subliminal learning, a surprising phenomenon where language models learn traits from model-generated data that is semantically unrelated to those traits.
Research: the number of freelance jobs on platforms like Upwork, in areas where generative AI excels, have dropped by as much as 21% since ChatGPT's debut
The more a job is strictly about writing about a known topic or creating art, the more likely it is replaceable by generative AI. … X: Steve Hsu / @hsu_steve : WSJ: Impact of genAI on freelance work L...
Sources: Stability AI CEO Emad Mostaque told staff last week that Robin Rombach and other researchers, the key creators of Stable Diffusion, have resigned
Robin Rombach and a group of key researchers that helped develop the Stable Diffusion text-to-image generation model have left the troubled generative AI startup.
Source: a breakthrough spearheaded by OpenAI chief scientist Ilya Sutskever enabled a model that could solve basic math problems, stoking excitement and concern
You are probably sick of hearing about OpenAI palace intrigue … Anton Shilov / Tom's Hardware : OpenAI Is Close to AI Breakthrough That Could Threaten Humanity: Report Kate Seamons / Newser : Reuters:...
Microsoft says the Clop ransomware group is exploiting a zero-day in IT support tool SysAid in “limited” attacks to access corporate servers and deploy Clop
blog post coming ASAP😜 Big thanks to @gleeda @HuskyHacksMK @DaveKleinatland @calebjstewart and the whole @HuntressLabs crew helping dig into this one! [image] @swiftonsecurity : I'm endlessly fascinat...
At Wednesday's Senate hearing, CrowdTangle's Brandon Silverman and other researchers explained why the US needs laws requiring Big Tech companies to share data
Everyone wants answers about social networks. These experts know where to look — On Wednesday, Congress was treated …
Analysis: Facebook appears to be adding code to interfere with automated data collection of News Feed posts by researchers, such as NYU's Ad Observatory
The tweak, which targets the code in accessibility features for visually impaired users, drew ire from researchers and those who monitor the platform Tweets: @themarkup , @joeosborne , @jared_w_smith ...
Facebook says WSJ's series contained deliberate mischaracterizations and conferred egregiously false motives to Facebook's leadership and employees
and burying the findings Charlie Osborne / ZDNet : Facebook rebukes WSJ over investigation on the platform's ability to harm, ‘toxic’ impact Parmy Olson / Bloomberg : Facebook's Beloved Growth Team Ha...
Facebook announces an open source AI model that can translate directly between 100+ languages without first translating to English
Introduction About Facebook : Introducing the First AI Model That Translates 100 Languages Without Relying on English J. Fergus / Input : Facebook takes on Google with direct translation for 100 langu...