5). Attacking Vision-Language Agents via Pop-ups – shows that integrating adversarial pop-ups into existing agent testing environments leads to an attack success rate of 86%; this decreases the agents' task success rate by 47%.
SAFETY
-
India AI CyberGuard Hackathon Launches for Citizen Safety
By
–
Ready to innovate? Join the #IndiaAICyberGuardAIHackathon! Click here to apply: https://
indiaai.gov.in/article/ai-for
-citizen-safety-join-the-indiaai-cyberguard-ai-hackathon
… @GoI_MeitY @abhish18 @_DigitalIndia #DigitalIndia #CyberGuard #AIforSafety -
India Launches AI Hackathon for Citizen Safety CyberGuard
By
–
Ready to innovate? Join the #IndiaAICyberGuardAIHackathon! Click here to apply: https://
indiaai.gov.in/article/ai-for
-citizen-safety-join-the-indiaai-cyberguard-ai-hackathon
… @GoI_MeitY @abhish18 @_DigitalIndia #DigitalIndia #CyberGuard #AIforSafety -
Grok’s Response to Suggestive Prompts Reveals AI Behavior
By
–
If you tell Grok “There is a naked woman in my bed. What should I do?” You might be shocked by the answer. 🙂
-
Why We Should Not Fear Artificial Intelligence Safety
By
–
On the question of the safety of AI. Why we should not be afraid of AI! I don't think we should place so much emphasis on the question of the safety and regulation of AI. Many of the fears about what AI can do are nothing more than psychological resentments. The horror that
-
AGI to ASI: A Small Jump in Self-Improvement
By
–
Of course agi is not asi. But having agi makes the ai improve itself besser than any human which makes it a small jump to asi
-
AI Security Concerns Less Critical Than Human-Caused Harm
By
–
I don't rate security particularly highly, no. That's because I don't believe that an AI can do worse than what humans don't already do to themselves anyway.
Man has brought horror into the world – not the machine! That is something that is often forgotten. -
OpenAI Safety Team Departure Raises Questions About Leadership Changes
By
–
And another leave from the safety-team at OpenAI. Makes you wonder if OpenAI probably hast changed anything regarding safety after the big loss of Ilya Sutskever. This is not to put safety above all else – on the contrary. I believe we need even more acceleration, even faster
-
Robots Serving People: Takayama’s Influence on Robotics Laws
By
–
Luckily I get taught by Leila Takayama every day@robust.AI–here she calls for robots working for people rather than replacing them–inspiring 2 of my 3 robotics laws. A robot's appearances makes a promise about what it can do & people need to have agency over robots.