Welcome to our monthly newsletter, which we have prepared to bring you current developments and informative content on Artificial Intelligence Security!
🙋🏻♀️Editor's Note:
Our website aisafetyturkiye.org is under construction. During this period, you can reach us via aisafetyturkiye@gmail.com.
3 months with mentors in the field of Artificial Intelligence Security. This interdisciplinary program, which will be conducted throughout the course, is designed for researchers studying complex and intelligent behavior in natural and social systems or working in fields such as mathematics, philosophy and engineering.
For anyone interested in the mission of making AI systems safe and useful, the date and location will be announced soon!
This initiative, which aims to accelerate Artificial Intelligence Security research and improve research capabilities, will provide funding and mentorship to 10-15 scholars to work full-time on Artificial Intelligence Security research.
Anthropic's Sam Marks describes providing guidance for AI researchers who want to reduce disaster risk but don't know where to start, and outlines key technical areas where advances today will ensure future AI systems remain safe and controllable.
OpenAI released its latest reasoning model o3; This model broke records in all comparison tests and amazed many experts in the world of artificial intelligence. So what does this mean for the path to artificial general intelligence?
Geoffrey Hinton, a pioneer in the field of artificial intelligence and winner of the Nobel Prize in Physics, increased the risk of human extinction due to artificial intelligence to 10-20% in the next 30 years.
Hinton is concerned about the rapid development of artificial intelligence and the difficulty of controlling systems superior to human intelligence (ASI).
The pace of AI development has exceeded Hinton's original predictions, and she expects systems that surpass human intelligence to emerge within about 20 years.
Hinton notes that examples of less intelligent beings controlling more intelligent ones are rare in nature, and the human-AI relationship poses a unique challenge.
Hinton advocates for government intervention to ensure the safe development of AI, arguing that companies alone will not be enough.
OpenAI's o3 system achieved unprecedented performance in the ARC-AGI-1 benchmark, scoring 75.7% under standard computational constraints and 87.5% with extended resources.
The o3 system represents a critical step in the progress towards artificial general intelligence (AGI) and has important implications for the future of artificial intelligence.
OpenAI has expanded access to its advanced text-to-video AI model Sora to ChatGPT Plus and Pro subscribers.
The move puts OpenAI in competition with established players like Meta, Google and Stability AI in the multi-modal AI space.
The new version of Sora offers features such as 1080p resolution, 20-second video creation and support for various aspect ratios.
OpenAI has implemented security measures such as content filtering systems targeting abusive material, anti-deepfake protocols, and controlled human subject processing.
