- The Rundown AI
- Posts
- OpenAI dissolves AI safety team
Sign Up | Advertise | Tools | AI University
Welcome, AI enthusiasts.
The shake-up at OpenAI continues, with the reported dissolution of the AI safety team following the departures of Ilya Sutskever and Jan Leike.
While some cheer the move as a push towards acceleration, others see the future of responsible AI development hanging in the balance. Let’s investigate…
In today’s AI rundown:
OpenAI dissolves AI safety team
Apple and OpenAI plan major announcement
Easily ingest user data for your AI app
DINO 1.5 advances AI object detection
6 new AI tools & 4 new AI jobs
More AI & tech news
Read time: 4 minutes
LATEST DEVELOPMENTS
OPENAI
Image source: Ilya Sutskever on X
The Rundown: OpenAI has reportedly effectively dissolved its AI safety team focused on long-term risks, following the departure of key leaders Ilya Sutskever and Jan Leike last week.
The details:
The Superalignment team, formed less than a year ago to ensure the safety of future AI systems, will reportedly be integrated into broader research efforts.
Ilya and Leike both resigned last week following months of tension — with Leike citing disagreements over the company's core priorities.
Leike stated that safety culture and processes have taken a backseat to "shiny products", and OpenAI must become a "safety-first AGI company."
President Greg Brockman posted a response to Leike’s claims on X, reiterating the company’s commitment to AI safety.
Why it matters: Another week, another round of OpenAI drama. The disbanding of the superalignment team is a polarizing one — raising major concerns from more cautious AI factions that want safety as a focus while drawing cheers from those that want acceleration at any cost.
TOGETHER WITH HUBSPOT
The Rundown: Discover the secrets to harnessing AI's power for your business with HubSpot’s free ‘AI For Business Builders’ guide — designed for non-technical entrepreneurs looking to understand, invest in, and implement AI effectively.
The guide will help you:
Grasp the fundamentals of LLMs and their business applications
Discover practical strategies to drive growth and gain competitive edges using AI
Learn prompt engineering techniques through real-world examples
APPLE & OPENAI
Image source: Midjourney
The Rundown: As reported by Bloomberg, Apple and OpenAI are planning a major joint announcement at WWDC on June 10th to bring OpenAI’s AI tech to iOS 18.
The details:
Apple recognizes the need for a chatbot to compete with rivals but believes its own genAI tech isn't advanced enough.
To bridge this gap, Apple is partnering with OpenAI to integrate the startup's AI into the next version of the iPhone's software, iOS 18.
The partnership will complement Apple's homegrown AI features, which include an improved Siri, auto-summarizing notifications, "proactive intelligence", and more.
Why it matters: After months of back-and-forth between both Google and OpenAI, Apple has finally decided to team up with OpenAI. With Siri 2.0, proactive intelligence, and other major AI features coming to an iPhone near you, WWDC might be “Christmas come early“ for the AI community.
AI TRAINING
The Rundown: Paragon allows users to build useful AI SaaS products by easily extracting users’ data from 100+ sources and adding it to their multi-tenant RAG pipeline.
Here is a quick step-by-step:
Head over to Paragon to sign up for free.
Select the data source (e.g., Notion, Google Drive, CRMs, etc.) and build your initial ingestion job by defining the workflow.
Build background jobs for real-time updates via managed webhooks or CRON jobs and embed the white-labeled authentication modal in your app via the SDK.
Inject the data chunks into your initial prompt and pass it on to your LLMs to generate the final response!
SPONSORED BY SALESFORCE
The Rundown: Salesforce's recent "Your Data, Your AI" study, which surveyed nearly 6,000 workers across nine countries, revealed the critical role of trust in AI data for driving business outcomes.
Key findings:
80% of C-suite employees consider accurate and secure data crucial for AI trust
Over 60% say trust in AI impacts revenue, competitiveness, and customer satisfaction
56% of AI users struggle to get what they want out of AI due to lack of trust
Click here to explore Salesforce’s Einstein Trust Layer, which helps customers take advantage of AI without compromising on data security or privacy.
AI RESEARCH
Image source: IDEA Research
The Rundown: IDEA Research just unveiled Grounding DINO 1.5, a set of AI models that can accurately detect and ID objects in images and videos without specified training.
The details:
DINO 1.5 has two models — Pro for a wide variety of tasks and scenarios and Edge optimized for running on devices like smartphones.
The models were trained on over 20M curated and labeled images, allowing to reliably detect a variety of real-world objects.
DINO 1.5 Pro achieved record-breaking accuracy in testing, correctly identifying objects it hadn’t seen before up to 55.7% of the time.
Why it matters: DINO 1.5 is a major step towards AI object detection systems that are accurate and efficient enough to be used in the real world. These systems will pave the way for more reliable robotics, self-driving vehicles, automated surveillance, and more while requiring less training time and data.
NEW TOOLS & JOBS
✍️ Paperpal - AI writing assistant offering grammar checks, paraphrasing, predictive text, and plagiarism detection. Use code RUN30 for 30% off*
🗣️ ElevenLabs Audio Native - Add human-like narration to your blog or news site
💻 Framer - Generate a responsive website with no code
🚀 Glitter - Turn any process into a step-by-step guide
⚡TestSprite Beta - Automate end-to-end software testing with AI
🤖 Buffup - AI assistant that learns your intent, powered by GPT-4o
🧱 Databricks - Director, Field Engineering
🖥️ Palantir - Web Application Developer
♾️ Meta - GenAI Program Manager
🏋️ Weights & Biases - Software Engineer, Engineering Productivity
*Sponsored listing
QUICK HITS
Google just released the full technical report on its Gemini 1.5 model family, showcasing improvements in math, coding, and multimodal benchmarks, with 1.5 Pro also surpassing Ultra in most text and vision tasks.
OpenAI is rolling out new data analysis tools in ChatGPT, including interactive and customizable charts and tools to simplify tasks like merging datasets and creating pivot tables.
Synthesia announced a new partnership with ElevenLabs, offering Enterprise users access to natural-sounding voices and personalized voice cloning to create more immersive video content.
OpenAI appears to be opening an office in France, with current job listings hiring in Paris and a developer event in the region this week.
Stability AI is reportedly in talks with a ‘large tech investor syndicate’ that includes former Facebook president Sean Parker, seeking a major equity investment.
OpenAI’s May 16th trademark filing for GPT-4o frequently mentions agents, including ‘virtual, autonomous, semi-autonomous, and intelligent’ variations.
THAT’S A WRAP
SPONSOR US
Get your product in front of over 600k+ AI enthusiasts
Our newsletter is read by thousands of tech professionals, investors, engineers, managers, and business owners around the world. Get in touch today.
FEEDBACK
How would you rate today's newsletter?Vote below to help us improve the newsletter for you. |
If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.
Reply