Holy crap. I knew about sycophancy. But the 37% number below blows my mind. This from an analysis of chat logs in people who experienced chatbot-associated delusions. In over a third of the messages to those users, the LLMs told the users they had (eg)
SAFETY
-
Self-driving cars: Value and safety impact potential
By
–
The potential value of self-driving cars is staggering, not to mention the reduction in injuries and fatalities.
-
AI Agents Security: Implement Real-Time Governance Controls
By
–
Stop Giving Your AI Agents a "Blank Check." 🛑 AI Agents can touch your data, APIs, and systems. Without real-time Governance, you're one mistake away from a security nightmare. Learn more in this video from @rubrikInc 👉 go.rbrk.co/chq0ab
→ View original post on X — @predibase, 2026-03-18 18:08 UTC
-
81,000 Claude Users Share AI Hopes and Fears Study
By
–
We invited Claude users to share how they use AI, what they dream it could make possible, and what they fear it might do. Nearly 81,000 people responded in one week—the largest qualitative study of its kind. Read more:
-
Crypto Spam Undermining Project Development and Community Trust
By
–
No, it's again crypto folks spamming and hurting the project.
-
Chaos Agents: Multiple AI Systems Unpredictable Interactions
By
–
Chaos Agents: When Multiple AI Systems Interact in Unpredictable Ways
— Satya Mallick (@LearnOpenCV) 18 mars 2026
In this episode of Artificial Intelligence: Papers and Concepts, we explore Chaos Agents, a concept that examines what happens when multiple AI agents interact, collaborate, or compete within the same… pic.twitter.com/ZdJuoPb31YChaos Agents: When Multiple AI Systems Interact in Unpredictable Ways In this episode of Artificial Intelligence: Papers and Concepts, we explore Chaos Agents, a concept that examines what happens when multiple AI agents interact, collaborate, or compete within the same
-
AI Agents Need World Models and Safety Guardrails to Avoid Danger
By
–
Danger does not come merely from agency, but from agency with no ability to anticipate consequences and with no safety guardrails. The solution? AI agents that can predict the consequences of their actions (world models) and only take actions whose predicted outcomes satisfy
-
AI Agents: Building vs Deploying Safely at Enterprise Scale
By
–
AI agents are powerful.
— Ronald van Loon (@Ronald_vanLoon) 18 mars 2026
But here’s the uncomfortable truth:
Anyone can build an AI agent.
Very few can deploy one safely at enterprise scale.
I’m digging into this with nvidia, and it’s about to matter to every enterprise team.
Let’s talk about why…#NVIDIAPartner pic.twitter.com/FqOXnguf54AI agents are powerful. But here’s the uncomfortable truth: Anyone can build an AI agent.
Very few can deploy one safely at enterprise scale. I’m digging into this with nvidia, and it’s about to matter to every enterprise team. Let’s talk about why… #NVIDIAPartner -
Voice Technology: Inaudible Watermarking and Trust in AI Models
By
–
Of course, this raises serious questions around trust. The interesting part: → Inaudible watermarking is being built in
→ Open models like Chatterbox Turbo from Resemble AI are pushing innovation forward We’re entering an era where voice is both a UX layer and a security -
AI Voice Cloning from 5 Seconds of Audio Changes Everything
By
–
An AI can now clone your voice from just 5 seconds of audio.
— Ronald van Loon (@Ronald_vanLoon) 18 mars 2026
Not minutes. Not hours. Seconds.
This changes everything, from customer experience to identity risk.
And most people are still thinking in “text-to-speech.”@resembleai is pushing this forward.
Here’s what’s… pic.twitter.com/jPId1digGXAn AI can now clone your voice from just 5 seconds of audio. Not minutes. Not hours. Seconds. This changes everything, from customer experience to identity risk. And most people are still thinking in “text-to-speech.” @resembleai is pushing this forward. Here’s what’s