Who is Andrew Tulloch? Mira Murati's co-founder who rejected a $1.5 billion payday from Mark Zuckerberg
However, the chief target for Zuckerberg and Alexandr Wang, Meta's newly appointed head of Superintelligence Labs, was Murati's co-founder Andrew Tulloch. Reportedly, Zuckerberg and Wang peppered Tulloch with messages asking him to join and offered a $1 billion pay package. Tulloch could have made as much as $1.5 billion, including top bonuses and stock performance, over a period of six years.
However, not only did Tulloch say no, none of Thinking Machines Lab's employees left either.
Originally from Australia, Tulloch is a co-founder at Thinking Machines Lab. While not much is known about the startup, Murati has raised over $2 billion from investors. The company's stated mission is to make 'AI systems more widely understood, customizable and generally capable.'.
Murati has said that the startup will share its first product in the 'next couple months'. It has reportedly leased an office in a quieter part of San Francisco's Mission District, a couple of blocks from OpenAI.
Tulloch graduated from the University of Sydney and had the highest GPA among science students. After that, he worked on machine learning at Facebook for 18 months before starting graduate school at Cambridge.
He eventually went on to work at Facebook's AI Research Group and became a distinguished engineer. OpenAI President Greg Brockman tried to hire him in 2016 as one of the company's first employees, but he was worried about taking a large pay cut from his $800,000 job at Facebook and decided not to join the startup. OpenAI was paying its new recruits an annual salary of $175,000 and a $125,000 annual bonus.
Tulloch did go on to join OpenAI in 2023 after ChatGPT had already become a viral sensation.
As per a report by Wired, Mira Murati's AI startup Thinking Machines Lab was offered $1 billion by Mark Zuckerberg's Meta to join its AI Superintelligence team.
'So far at Thinking Machines Lab, not a single person has taken the offer,' Murati was quoted as saying by the outlet.
According to the report, Mira Murati's team members were offered packages that ranged between $200,000 to $1 billion, every single one of which was rejected as employees refused to join the Meta Superintelligence Lab.
Notably, Zuckerberg has been trying to revive Meta's AI efforts in recent months after the company's Llama 4 models failed to gain widespread recognition compared to rival models from OpenAI, Google and Anthropic. The Meta CEO has set up a new Superintelligence Labs unit in the company to aim for 'superintilligence' — the cognitive performance of the brightest and most gifted human beings.
Meta has also acquired a 49 percent stake in AI startup Scale AI for $14.8 billion and AI-generated human voices startup Play AI. The company has also tried to poach hundreds of OpenAI researchers and brought at least 10 of them onboard along with many others from Apple, Google and Anthropic.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
India Today
11 minutes ago
- India Today
Mark Zuckerberg doubles offer to Rs 2,196 crore after 24-year-old AI researcher says no to Meta
Silicon Valley is blazing through the AI talent wars, and Meta is clearly sparing no expense to attract the brightest minds in technology to build its Superintelligence Lab. According to reports Meta is offering hefty compensation packages to lure top talent, the latest being a staggering Rs 2,196 crore compensation, made personally by Meta CEO Mark Zuckerberg to 24-year-old AI researcher Matt to The New York Times, Deitke initially declined Meta's offer of approximately $125 million (around Rs 1,098 crore), spread over four years and chose to focus on his rapidly growing AI startup, Vercept. However, following his refusal, Zuckerberg personally arranged a face-to-face meeting and revised the offer to an extraordinary $250 million (around Rs 2,196 crore), including handing over up to $100 million (Rs 879 crore) in the first year alone. This final offer reportedly convinced Deitke to join Meta as an AI researcher within its Superintelligence who exactly is Matt Deitke, and why was Zuckerberg so determined to bring him into Meta's superintelligence project? Deitke is considered one of the most sought-after minds in the AI community. Originally a PhD candidate at the University of Washington, he left academia to pursue real-world AI breakthroughs at the Allen Institute for AI (AI2) in Seattle. There, he led the development of Molmo, which is a cutting-edge multimodal chatbot capable of processing not only text but also images and audio. His work for Molmo earned him an Outstanding Paper Award at NeurIPS 2022, one of the most prestigious conferences in the AI Deitkework in the AI world, Meta had been keen to recruit him since late 2023, but he initially turned down their lucrative offer in favour of building his autonomous AI agents at Vercept. However, after Zuckerberg's personal involvement, and the increased compensation package ultimately made Deitke decide to join Meta's ambitious Superintelligence Meta's Superintelligence lab has already invested over 1 billion to build an 'all-star' AI team, recruiting top talent from rivals like OpenAI, Google, and Apple. One of its latest high-profile hires along with Deitke is Ruoming Pang, the former head of Apple's AI models team, reportedly offered a compensation package exceeding $ 200 million. - Ends
Time of India
11 minutes ago
- Time of India
OpenAI, Google and Anthropic win US approval for civilian AI contracts
Academy Empower your mind, elevate your skills The US government's central purchasing arm is adding OpenAI , Alphabet's Google and Anthropic to a list of approved artificial intelligence vendors, opening the door to widespread adoption of the technology across civilian federal move by the General Services Administration, to be announced Tuesday, will speed up the adoption of AI tools in the federal government by making them available through its Multiple Award Schedule, a federal contracting platform with contract terms already set. Without that flexibility, agencies would ordinarily spend months negotiating their own terms for use of the officials said the models from the three companies — OpenAI's ChatGPT , Google's Gemini and Anthropic's Claude — were evaluated by several performance and security agency didn't immediately disclose the terms of the contracts. But it's used its buying power to negotiate deep discounts with software providers like Adobe Inc., Salesforce Inc. and leading-edge AI companies would also be considered for the marketplace. The first three vendors were simply further along in the procurement process, the officials said.'We're not in the position of picking winners or losers here. We want the maximum number of tools to provide to all federal government employees to make them as productive as possible,' said GSA Deputy Administrator Stephen Ehikian. 'There's going to be different tools for different use cases.'The move comes just days after President Donald Trump signed three executive orders aimed at reshaping the government's role in AI, including a mandate that federal agencies only procure language models 'free from ideological bias.'Enforcing the presidential ban on what Trump calls 'woke AI' would be an agency-by-agency process, according to the GSA.'But at the same time, this is a race, right? And as the president said, we're going to win this race,' said Josh Gruenbaum, commissioner of GSA's Federal Acquisition the three companies to the multiple award schedule — which makes commercial IT products more readily available to agencies — means federal bureaucrats can begin using large language models that had previously been restricted to smaller pilot projects or national security use. The Pentagon has already awarded AI contracts to OpenAI and Elon Musk's xAI, which are separate from the GSA's announcements on agencies, including the Treasury Department and Office of Personnel Management, have already expressed interest in using the new platform, according to GSA officials. Under the previous presidential administration, federal agencies identified potential uses for AI such as processing patent applications, detecting tax fraud, reviewing grant submissions and copy-editing press of Personnel Management Director Scott Kupor said he envisions using AI to develop customer service chatbots and to summarize tens of thousands of public comments on rulemaking — a process that previously bogged down changes to he said agencies will also have to hire savvy employees. 'We're probably missing people who are super conversant with very modern, AI-related stuff,' he said.'Clearly, we can't just throw things against the wall and see what sticks,' Kupor said.
Hans India
an hour ago
- Hans India
OpenAI Set to Launch GPT-5: A Faster, Smarter, and More Capable AI Model
This highly anticipated update to the GPT series promises significant upgrades, including faster performance, improved reasoning, and enhanced multimodal capabilities. The model is also expected to support autonomous task execution, potentially changing how users interact with artificial intelligence. Confirmation of GPT-5's upcoming launch came from none other than OpenAI CEO Sam Altman. He teased the release on X (formerly Twitter), saying the company is 'releasing GPT-5 soon.' During a recent appearance on the This Past Weekend podcast with comedian Theo Von, Altman elaborated on his experience with GPT-5's abilities. Describing the model as 'very intelligent,' Altman recounted a moment where GPT-5 solved a problem he couldn't crack himself. 'I put it in the model, this is GPT-5, and it answered it perfectly,' he said, adding, 'It was a weird feeling—I felt useless relative to the AI.' Although specific technical details remain under wraps, OpenAI has hinted that GPT-5 will not be a routine update. Instead, it will be a major leap forward by combining several of the company's existing technologies into one unified model. This integration could eliminate the current need for users to switch between different models based on their task—such as reasoning, real-time interaction, or multimodal processing. Here's what's expected in GPT-5: Advanced Reasoning: Industry insiders suggest that GPT-5 might demonstrate near PhD-level proficiency in logic, problem-solving, and complex cognitive tasks. Industry insiders suggest that GPT-5 might demonstrate near PhD-level proficiency in logic, problem-solving, and complex cognitive tasks. Multimodal Expansion: Building on GPT-4o's voice and image functions, GPT-5 could introduce video processing capabilities, allowing users to interact via text, voice, images, and video—seamlessly. Building on GPT-4o's voice and image functions, GPT-5 could introduce video processing capabilities, allowing users to interact via text, voice, images, and video—seamlessly. Massive Memory Boost: GPT-5 is expected to feature a context window exceeding 256,000 tokens, twice that of GPT-4o. This would enable significantly longer and more coherent conversations, while also improving continuity between sessions. GPT-5 is expected to feature a context window exceeding 256,000 tokens, twice that of GPT-4o. This would enable significantly longer and more coherent conversations, while also improving continuity between sessions. Autonomous Task Execution: One of the most anticipated advancements, GPT-5 could allow AI to independently complete complex, multi-step tasks with minimal user input—paving the way for intelligent agents. Sam Altman has praised these enhancements during internal testing, calling GPT-5 'very fast' and admitting it made him feel 'nervous' about the growing capability of AI. In terms of availability, GPT-5 will reportedly come in three versions: A flagship model for standard use, A mini version optimized for lightweight tasks within ChatGPT, And a nano version aimed at developers and enterprise users through the OpenAI API. Altman also confirmed that OpenAI plans to release more models, products, and features throughout the year, making 2025 a landmark for AI innovation. As excitement builds, GPT-5 is poised to redefine the boundaries of what AI can achieve—ushering in a new era of smarter, more adaptive, and truly multimodal tools for work, creativity, and problem-solving.
