Research: https://
anthropic.com/research/many-
shot-jailbreaking
…
SAFETY
-
Anthropic Research: Many-Shot Jailbreaking Techniques Analysis
By
–
-
Repetition Jailbreak: Context Window Vulnerability in Large Language Models
By
–
New jailbreaking technique: pure repetition. AIs are getting big context windows, it turns out if you fill a lot of it with examples of bad behavior, the AI becomes much more willing to breach its own guardrails. Security people are used to rules-based systems. This is weirder.
-
LLMs Risk: Hallucinations and False Source Attribution
By
–
I think that's OK: it's an analogy. Bears could kill you. LLMs could embarrass you by causing you to cite a non-existent source.
-
LLM Hallucinations in Medical Assessments: A Critical Concern
By
–
In a recent study, 4 out of 5 LLMs hallucinated a significant proportion of sources for their answers to medical questions. Should we be using them for medical assessments?
-
Three Laws of Robotics and 50 Million Robots Sold
By
–
My three laws of robotics. The companies that I have founded have sold 50+ million real robots to real customers; research robots, home cleaning, nuclear power plant inspection, military ground robots, upper body humanoids in factories, and now at http://
Robust.AI, -
Watch Out for Sketchy AI Strategies and Tactics
By
–
There are going to be some sketchy strategies out there, so watch out. https://t.co/X75pBslY6L
— Ethan Mollick (@emollick) 2 avril 2024There are going to be some sketchy strategies out there, so watch out.
-
Sam Altman: GPT-5 Ready but Too Risky to Release
By
–
“GPT-5 is ready, but it’s too risky to release”, says Sam Altman. “In the meantime, let’s dive into the treasure trove of bustling artificial intelligence landscape.”
-
First they came for epistemology: AI impact on knowledge
By
–
Epic. "First they came for the epistemology. We don't know what happened after that" is so catchy
-
Stanford Launches Center for Research on Foundation Models
By
–
August 2021: Recognizing a paradigm shift in AI, we launched the Center for Research on Foundation Models (
@StanfordCRFM
) led by @percyliang and published a groundbreaking report on the opportunities and risks of foundation models. #MondayMilestones 11/n -
Two Delivery Robots Deadlocked in Standoff Situation
By
–
Two delivery #Robots refusing to give way to each other
— Ronald van Loon (@Ronald_vanLoon) 1 avril 2024
via @FrRonconi#AI #Robotics #FutureOfWork #Tech #Technology #Innovation
cc: @space_mog @pbalakrishnarao @chr1sa pic.twitter.com/EFkZrDRFvfTwo delivery #Robots refusing to give way to each other
via @FrRonconi #AI #Robotics #FutureOfWork #Tech #Technology #Innovation cc: @space_mog @pbalakrishnarao @chr1sa