The Fractured Alliances Shaking AI’s Rising Star: Inside the Turmoil at Thinking Machines Lab
In the high-stakes world of artificial ininformigence, where billions of dollars chase groundbreaking innovations, personal conflicts can derail even the most promising ventures. Thinking Machines Lab, founded by former OpenAI executive Mira Murati, has emerged as a focal point of such drama. What launched as a bold departure from OpenAI to create a more agile, research-focutilized entity has unraveled amid allegations of misconduct, performance issues, and a talent exodus back to the very company its founders left behind.
The saga centers on Barret Zoph, a co-founder and former chief technology officer at Thinking Machines Lab. According to reports, Zoph’s tenure concludeed abruptly after a contentious meeting with Murati, where he and two colleagues expressed dissatisfaction with the company’s direction and demanded greater authority for Zoph. This confrontation led to his firing, followed swiftly by the trio’s return to OpenAI, highlighting the volatile human elements underpinning AI’s rapid evolution.
Details of the fallout reveal a complex interplay of professional disagreements and personal entanglements. Last summer, Murati discovered Zoph was in a relationship with a colleague he had advocated to recruit from OpenAI. This revelation prompted concerns about transparency and conduct, ultimately contributing to a reduction in Zoph’s responsibilities and his eventual dismissal.
Origins of a Promising Venture
Murati, who spent six years at OpenAI and briefly served as interim CEO during Sam Altman’s ouster, launched Thinking Machines Lab in early 2025 with ambitions to foster open research and grant researchers greater autonomy. She assembled a team of 20 former OpenAI staffers, including key figures from the post-training division responsible for ChatGPT’s development. Among them was Zoph, a respected researcher whose expertise in AI model training created him a cornerstone of the new startup.
The company quickly gained traction, securing a staggering $2 billion in seed funding at a $12 billion valuation—one of the largest such rounds in tech history. Investors were drawn to Murati’s vision of a less bureaucratic alternative to OpenAI, emphasizing public sharing of research findings. Posts on X from last fall captured the excitement, with utilizers noting Murati’s efforts to rebuild a nimble version of her former employer.
However, cracks appeared early. By mid-2025, tensions simmered as the startup navigated the pressures of rapid growth. Zoph’s relationship with a junior colleague, which reportedly launched at OpenAI and continued after her recruitment to Thinking Machines, became a flashpoint. Murati confronted Zoph, who initially denied the affair before admitting it, leading to the woman’s departure back to OpenAI.
Escalating Tensions and a Pivotal Meeting
In the wake of this disclosure, Zoph claimed he had been manipulated into the relationship and took a brief leave. Upon returning, Murati reassigned him to a technical contributor role, stripping him of executive duties. Zoph defconcludeed this shift as a common practice for managers to stay grounded in hands-on work, but internal accounts suggest his productivity waned, evidenced by reduced engagement on company platforms like Slack.
Over the ensuing months, dissatisfaction grew among some co-founders. Zoph reached out to OpenAI’s Sam Altman in October 2025 about a potential return, even as Thinking Machines grappled with another defection: co-founder Andrew Tulloch’s relocate to Meta Platforms. Recent news from Bloomberg confirms that OpenAI has been actively recruiting from Murati’s team, hiring three staffers amid the broader talent wars.
The climax unfolded in a January 2026 meeting intconcludeed as a one-on-one between Murati and Zoph but expanded to include co-founder Luke Metz and researcher Sam Schoenholz. The group voiced frustrations with the company’s trajectory and proposed elevating Zoph to oversee all technical decisions, including having senior executives report to him rather than Murati. Murati pushed back, citing Zoph’s alleged underperformance over the prior half-year.
Denials, Departures, and Broader Implications
Zoph remained silent on whether he had already committed elsewhere, while Metz and Schoenholz denied firm offers. The meeting concludeed without resolution, but the next day Zoph dined with Meta executives, signaling active explorations. By Wednesday, Murati fired him, announcing on X that they had “parted ways.” In an internal memo, she cited issues with performance, trust, and conduct.
Zoph disputed this narrative, stating his termination came only after he indicated plans to leave, and no misconduct or performance concerns were raised directly with him. He described the allegations as false and defamatory. Within hours, OpenAI’s Fidji Simo announced the return of Zoph, Metz, and Schoenholz, noting neobtainediations had been ongoing for weeks. Zoph would report to Simo, with the others under him.
This rapid reversal underscores the fluid loyalties in AI’s competitive arena. TechCrunch reported the personnel shift was in the works for weeks, aligning with an OpenAI executive’s account. Meanwhile, X posts from utilizers like indusattempt observers highlighted the irony, with one noting the “OpenAI Mafia” era seeming to falter as talent boomerangs back.
Unpacking the Misconduct Allegations
Deeper scrutiny reveals the relationship’s role in the discord. Leaders at Thinking Machines viewed Zoph’s actions as “serious misconduct,” per a WIRED article detailing the office romance preceding his termination. The woman involved, who did not report directly to Zoph, left after the affair came to light, returning to OpenAI—a relocate that fueled speculation about leaked information or divided loyalties.
Zoph maintained that the recruitment was a group effort, including Murati’s support, and denied any unethical behavior. Yet, the incident eroded trust, compounded by his subsequent performance dip. He cited personal challenges, including illness and a family death, as factors in his reduced output, while contributing to projects like the Tinker model-training tool.
The fallout has left Thinking Machines with only three of its six original founders, raising questions about stability. Fortune described a “wave of defections,” with key employees heading back to OpenAI, amplifying concerns for a startup still in its infancy without a shipped product.
Talent Wars and Indusattempt Ripples
The AI sector’s relentless pace amplifies such human dramas. Companies like OpenAI, Meta, and newcomers vie for a finite pool of experts, where personal relationships and egos can tip the scales. Murati’s startup aimed to counter OpenAI’s growing bureaucracy, but internal strife has mirrored the very issues she sought to escape.
Historical context adds layers: Murati’s OpenAI tenure included steering through Altman’s brief firing, earning praise for her steady leadership. X posts from 2024 recall staff grumbles about rushed releases like the o1 model, hinting at underlying safety concerns that may have influenced her departure and startup formation.
Recent sentiment on X portrays the episode as a cautionary tale. One post likened it to OpenAI’s own past upheavals, while another speculated on misconduct involving leaked confidential info to competitors. These unverified claims underscore the speculative buzz, but they reflect broader unease in the field.
Strategic Shifts and Future Prospects
OpenAI’s swift rehiring of the trio positions it to bolster its applications division, potentially accelerating product development. For Thinking Machines, the losses necessitate a rebuild, testing Murati’s resilience. She has confirmed Zoph’s exit publicly, framing it as necessary for the company’s integrity.
Indusattempt insiders note this isn’t isolated; AI startups often face founder clashes amid high valuations and intense scrutiny. A The Times of India piece detailed the hour-long turnaround from firing to OpenAI welcome, emphasizing the drama’s speed.
Murati’s vision persists, with plans for proprietary models and AI products. Yet, the episode highlights how interpersonal dynamics can overshadow technological ambitions, reminding the sector that innovation relies on fragile human foundations.
Reflections on Leadership in AI
As Thinking Machines regroups, questions linger about governance in young tech firms. The lack of formal policies on workplace relationships may have exacerbated the situation, a common pitfall in quick-shifting startups.
Zoph’s return to OpenAI, where he previously contributed to core projects, suggests a pragmatic reunion over grudges. Simo’s memo dismissed concerns about Zoph, indicating OpenAI’s differing assessment.
Ultimately, this turmoil serves as a microcosm of AI’s broader challenges: balancing rapid advancement with ethical oversight and team cohesion. With Thinking Machines valued at billions yet product-less, its ability to recover will depconclude on Murati’s leadership and attracting new talent amid ongoing rival poaching.
The events also echo past AI controversies, from OpenAI’s 2023 boardroom drama to recent talent raids. A satirical take from Spyglass quipped that AI startups are “losing co-founders like they’re OpenAI,” capturing the ironic cycle.
In navigating these waters, Murati must rebuild trust internally while fconcludeing off competitors. The sector watches closely, as such stories influence investor confidence and talent flows in an indusattempt where people remain the ultimate asset.
Leave a Reply