Title here
Summary here
🙋🏻♀️ Editor’s Note:
Have you noticed that we sent our newsletter 1 day later than regular this month? 👀
As of next month, we’ll continue on the second day of the second week of each month, as always!
ANNOUNCEMENTS 🔊
Cosmos Ventures Grant Program
This low-overhead grants program backs the next generation of thinkers and builders at the intersection of AI and human flourishing. Projects often take an interdisciplinary, humanistic approach, drawing from philosophy, computer science, political theory, economics, natural science, and other fields.
🗓️ Register by: November 1st
AI Safety Camp – Virtual 2025
A 3-month online research program where participants form teams to work on pre-selected projects.
🗓️Register by: November 17th
The Competition for LLM and Agent Safety
CLAS 2024, a NeurIPS competition, challenges participants to address critical AI safety issues. The event features three tracks: jailbreaking attacks, and backdoor trigger recovery for both models and agents. Winners share a $30,000 prize pool and the opportunity to co-author a publication on the competition results.
🗓️ Register by: October 18th
LASR Labs
LASR Labs offers a 3-month full-time, paid program in London focusing on technical AI Safety research. Participants work in small teams under experienced supervision to address concrete threat models and produce academic papers and blog posts on reducing risks from advanced AI.
🗓️ Register by: October 27th
TOP PICKS 📑 🎧
SB1047 is vetoed. Now what?
In our previous issue, we covered the highly impactful SB1047 bill regarding AI safety, which has been vetoed by the California state governor. What does this mean for the political perception of AI safety?
The AI Safety Institute Network: Who, What and How?
At the Seoul AI Safety Summit, US Secretary of Commerce Gina Raimondo announced a global network of AI Safety Institutes (AISIs), spanning from Kenya to the UK, Singapore to Canada. Alex Petropoulos from the International Center for Future Generations explores what AISIs are, their planned activities, and how they differ across countries.
NEWS 📰
OpenAI Transforms into a For-Profit Company
- OpenAI plans to restructure its primary business model and become a for-profit company.
- This decision comes after last year’s boardroom crisis, which resulted in Sam Altman’s temporary dismissal and subsequent return.
- The change will allow OpenAI to operate more like a traditional startup, though it also raises concerns about AI safety.
- For the first time, CEO Sam Altman will own shares in the company following the restructuring. The company’s valuation is expected to reach $150 billion.
Senior Executives Continue to Leave OpenAI
- OpenAI is experiencing significant changes in its leadership team with the departures of CTO Mira Murati, Chief Research Officer Bob McGrew, and Vice President of Research Barret Zoph.
- CEO Sam Altman stated that the resignations occurred amicably and were due to the intense workload associated with leadership roles.
OpenAI Unveils New AI Models
- On September 12, OpenAI introduced two new AI models, named “o1-preview” and “o1-mini.”
- These models utilize the “Chain of Thought” method, enabling them to think for longer durations and solve complex problems more effectively than previous models.
- The new models show significant improvements, particularly in core science, coding, and mathematics.
- For instance, according to OpenAI, the o1 model achieved an 83% success rate on International Mathematical Olympiad questions.
Council of Europe AI Treaty Signed
- On September 5, the Council of Europe opened for signing the “Framework Convention on Artificial Intelligence and Human Rights, Democracy, and the Rule of Law” (CETS No. 225) in Vilnius.
- This treaty is notable for being the first legally binding international agreement aimed at ensuring that the use of AI systems aligns with human rights, democracy, and the rule of law.
- It has been signed by numerous countries and organizations, including Andorra, Georgia, Iceland, Norway, Moldova, San Marino, the United Kingdom, Israel, the United States, and the European Union.
JOB POSTINGS 👩🏻💻
You can check out 80,000 Hours’ job board
to explore new opportunities in AI Safety!
Our October Meetup 🤍
At our AI Safety Türkiye October 2024, meetup we went over the online training sessions lined up for this year, the events we’ll be co-hosting with student groups, and the anticipated changes from the upcoming AI law in Turkey.
💌 We hope to see you at our next meeting!