Purported is the key word. The more of the face that’s covered the less accurate.
SAFETY
-
Facial Recognition Systems Enable Targeted Harassment Campaigns
By
–
What smug fog do people live in? How is this a gotcha? They’re covering their faces bc today, in 2024, facial rec systems exist that can easily ID them, IDs that are then used by coordinated actors (trustees funding moving billboards, etc) to doxx/violently intimidate them.
-
System Prompts Don’t Guarantee Truth: Manipulation Techniques
By
–
System prompt leaking techniques work, but just because something is stated in a system prompt doesn't mean that thing is actually true System prompts don't have to tell the truth, their job to influence the model to behave in certain ways
-
LLMs Structured Output as Postel’s Law Implementation
By
–
LLMs that implement constrained sampling/schema respecting structured output seem the ultimate implementation of Postel's Law: "Be conservative in what you send, be liberal in what you accept." Basically the ultimate interface block between systems.
-
Improving AI Policy Transparency for Model Users
By
–
I like that it's at least documented as a policy on https://
lmsys.org/blog/2024-03-0
1-policy/#our-policy
… – a small improvement could be provide a link to that policy that shows up when you're interacting with a model covered by it, to make things less confusing and opaque -
Med-Gemini-M Analyzes Laparoscopic Surgery Video for Safety Assessment
By
–
Here's one example of assessing a Med-Gemini-M 1.5 analyzing a video clip from the Cholec80 dataset to assess achievement of the Critical View of Safety (CVS) during a laparoscopic cholecystectomy.
-
Responsible AI: Compliance and Security Tactics Explained
By
–
Unlocking ResponsibleAI ’s Potential – Top Compliance and Security Tactics Watch Now: http://
bit.ly/AIGovSecurity Join me, Head of @Infosys #ResponsibleAI Syed Ahmed & host @Ronald_vanLoon as we cover: #AI 's Double-Edged Sword #MachineLearning Security -
Strong to Weak Alignment: A New Perspective on AI Safety
By
–
tired: weak to strong alignment wired: strong to weak alignment
-
AI Risks Repeating Web3’s Mistakes: Hype Over Utility
By
–
I watched Web3 drown in its own hype. Now I wonder if I'm seeing warning signs in AI. We're falling into the same traps: 1. Prioritizing flashy tech over real-world utility 2. Unclear terms (e.g. “AI agents”) 3. Building for other insiders rather than the mainstream
-
Models Cannot Accurately Answer Questions About Themselves
By
–
I don't trust models to answer questions about themselves accurately